Testing the 76

[M Hofbauer]

Vanir,

again I was just relating the findings of former discussions on this and the way I recollected them.

The only time I ever paid atention to tungsten use was during All or Nothing. When the german (censored) showed up (censored) in the game, I saw that my british (censored) first used regular AP with their 76, and the first round ricocheted off of the german (censored), and the second round was a tungsten which put the (censored) out of commission. So, for me it worked exactly as advertised, so I didn't worry further.

Your findings however indicate that it was just chance that it all worked out that way in that game. Also, those were british. Brits do funny stuff at times in general, and their 76/tungsten is quite different from the american one, so their tungsten attitude might be different.

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

"All i hear is the Iron Cross sucks etc. " (GAZ_NZ)

[Slapdragon]

Ok, I reran the series letting the Shermans plug away at the Panthers. In 10 range exercises of 10 tanks each at 490 meters:

2 turret penetrations kills

2 armament kills

12 surrendered

53 upper hull kills

29 abandoned tanks with no penetration

1tank, with 21ap left, sued a tungsten round I had forgoten to unload.

Interestingly enough, 500 meters is sometype of cut off, because my earlier tests at 520 meters resulted in no turret penetrations.

So, at 500 meters their is a 1-7% chance of killing a turret (with margin of error figured.

If you ever have several happen, you are the victim of collosal bad luck.

[Wolfe]

Hmmm. Tried switching the Panthers to King Tigers. And guess what type of round the Hellcats used at 500m?

Yup. Tungsten. 7 of the 8 Hellkitties chose Tungsten as their First Shot against the KTs. No AP. No bracketing. Just hell-bent on killing those KTs.

BTW, the hulldown KTs shows up as a 'Low' chance of a kill vs the Tungsten-laden Hellcats. It seems to me the 'OK' designation that a Hellcat (or Sherman 76) receives against the Panther is what might be preventing the use of Tungsten rounds. If the tank thinks AP will do (even when it likely won't), it's not going to waste its valuable Tungsten rounds.

- Chris

[PzKpfw 1]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Slapdragon:

Oh yes, the other solution: You are playing a different version of the game, or your game has been "hacked" somehow.

As it is every simulation done by three people cannot find supporting evidence for your claim. Your tanks were hit by something other than 76, the range was considerably shorter than 500 yards, or they hit something other than the turret face.

If you can send me a turn of a game that has a 76 killing a Panther at 500-700 rounds with a front turret strike then I will have to modify my position, as it is you are just mistaken for some reason: the 76 cannot reliably kill past around 350 meters unless it strikes some other part of the tank.<HR></BLOCKQUOTE>

Slapdragon, I just stated my experiences have been diferent with the 76 vs Panther then those, experienced by others here, accept it or don't it makes little diference to me, as all i'm doing is relaying what has happened to me.

Just don't sit their and tell me what my results etc, have to match yours or my CM is hacked etc, because of your empirical test's.

I play with Panther's rarely as I prefer other tanks, so when I use them its generaly when I let the AI choose my force or in a canned scen, next time it occurs I'll be sure to film it for you.

This is a program & many things can happen that can't be duplicated by tests I saw it all thru SP's, 2 & 3 & Pz. Elite etc, we had things that occured concerning penetration in SP that were never duplicated, Ie, over 27 75L70, 88L71, & 75L48 bounces off a Sherman at 700ms was one occurance, no one not even I with my ever bad luck with weird occurences, could duplicate it, & it happened to me, did that mean they never occured, nope, not according to Gary & Keith, they called it that 'random factor' which can occur even when its not suposed to.

I'll also go back to an discussion we had here a little while ago where a 76 Jumbo Sherman continued to use APCBC vs a Panther at 500ms, despite available APCR-T ammo, I pointed out that tac AI knew the 76mm APCBC could defeat the turret front armor at that range & thats prolly why it refused to fire APCR-T, Steve also commented that that very well could be the case of what was happening as the TAC AI won't use APCR-T if its normal ammo has an about even, even or better chance to penetrate.

Regards, John Waters

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

Notice: Spelling mistakes left in for people who need to correct others to make their life fulfilled.

[This message has been edited by PzKpfw 1 (edited 09-17-2000).]

[Slapdragon]

No problem: we were trying to empirically test the question. At work during research I get "war stories" all the time, so we test them empirically. We use observational data (ie: a game files showing a 76mm kill at 700 meters) or an some sort of test (ie: a firing range where the 700 meter kill can be observed). Vanir did observe a 500 meter turret kill, and at 100 tries I was getting like 2%.

My comments were more for those reading the post: War stories are passed around, like the old saw about the Stuart killing the King Tiger from 500 meters and how unfair that was. People should take war stories with a grain of salt. Scientific style tests though, the same types used to backwards engineer software or invent a new vaccine, with a data set and available, downloadable "proof" should always be taken to mean more than war stories.

[Vanir]

Wolfe:

I reran my earlier test using KTs instead of Panthers and there was no difference. Even though the Jumbo's chance to kill was "rare" they still never used a single tungsten round.

How many tungsten rounds and AP did your Hellcats have? TDs tend to carry a lot more tungsten than tanks and the ratio of tungsten to AP rounds seems to be a determining factor. I was using a typical tank loadout of 30 AP 30 HE 3 tungsten and 5 smoke. If anyone can get a unit to use tungsten at all with this type of loadout I'd like to know about it, because I sure can't.

I wonder about the kill chance ratings. At 700m the Jumbo gets a "ok" chance to kill a Panther even though they bounce dozens of shells off them with no penetrations. At the same range the chance is "rare" vs. KTs when I think the actual chance is "no freakin' way!" The only explanation I have for this is that the use of tungsten is factored into this. But this can be quite deceiving when the tank has no intention of ever using its tungsten.

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

No, there will be no sequels. Charles and Steve have given up wargame design in disgust and have gone off to Jamaica to invest their new-found wealth in the drug trade. -Michael emrys

[This message has been edited by Vanir (edited 09-18-2000).]

[Slapdragon]

In 134 shots the E-8s actually used Tungsten once: I had them loaded with 30 AP and 2 tungsten firing at 700 meters. Tank Destroyers may actually be more prone to use their Tungsten. I suspect that there are several variables involved in the question.

[Dschugaschwili]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Schugger:

After reading your extensive playtesting there is another point that concerns me:

The low ratio of front turret penetrations at ranges under 500m.

All panthers are rated with 85% armour quality which results in a front turret armour of 93,5mm at eleven degree.

Normaly those 76mm guns ( 793m/s) should have no problem in penetrating the Panther's turret at ranges of 500 meters.<HR></BLOCKQUOTE>

I don't know if the armor quality is multiplied by the armor thickness to get the "effective" armor thickness. Actually, I doubt it. It may as well indicate a higher probability of weak point penetrations or something totally different.

Dschugaschwili

[PzKpfw 1]

Awoke this AM with this in my box from Fionn who asked me to pass this info on as he is seeing some incorrect assumptions/conclusions on in this thread:

Slapdragon: I use 500 metres as my "good kill" range for a 76mm firing APCBC... In a recent game I nailed two Panthers with front turret penetrations at around this range. One was at 480metres, the other at 550 metres. These were both one shot kills. My first shot hit and killed the Panther. And the first shot was just a regular 76mm shot...

Anything below 350 to 250 is pretty much a "sure kill" whilst I don't like firing at less than this since I feel the 76mm APCBC is too likely to shatter when striking at the kinds of velocities it is travelling at below 200 to 250 metres.

Vanir: The AI evaluates 3 things when deciding whether to fire an AP or Tungsten

round.

1. Chance of hitting the enemy target.

2. Chance of an AP round penetrating said target.

3. Number of Tungsten rounds left.

So, if the chance of hitting the target is LOW then Tungsten won't be used unless there is literally no APCBC left.If an AP round has a good chance of hitting and killing the target then AP will be used and Tungsten saved.

If there are few tungsten rounds then APCBC will be used preferentially. If there are many Tungsten rounds then they will be used from the get go.

It is the INTERPLAY of these factors which determines when Tungsten is used and not just any one isolated factor.

Genaral Comment concerning Tungsten

As for the "Tungsten isn't supposed to be used as a first round" thing. It ISN'T IF the tank has a normal loadout. A normal loadout is one in which Tungsten is "rare" and so the AI uses AP shells to bracket the target.

OTOH if you skew the normal loadout and add more Tungsten then the AI says, "Hell, I have lots of Tungsten and this is a big enemy tank... I have enough Tungsten to waste 1 or 2 on bracketing shots ( in the hope that one of the braketing shots will hit and kill the tank."

Again, you can't simply take a statement which is obviously made when referring to normal loadouts and then apply it to an abnormal loadout and then wonder why it no longer applies.

Schugger: You can't apply the armour quality to the armour using such a simple

formula( Armour thickness multiplied by the armour quality divided by 100).. Thus

your point is incorrect. Armour quality is more a measure of flaking chances

etc and not just a straight 100mm of 80% quality = 80mm of 100% quality.

General Comments:

You should all listen to Markus and John more. They know what they're talking about as regards this stuff.

Personally I think most people here are forgetting about the extra armour thickness introduced when firing at an enemy tank whose turret isn't turned fully towards one. I wouldn't ever dream of taking a shot at 500 metres at a Panther whose turret was turned 30 degrees away from my gun. If the Panther's gun is pointed straight at me than a 500 metre shot is an easy kill for a 76mm. I think people may not be aware of this and certainly aren't accounting for it. Again, it is simply a case of people not really accounting for every factor and thus achieving incorrect results.

End....

Regards, John Waters

[This message has been edited by PzKpfw 1 (edited 09-18-2000).]

[Jeff Heidman]

I am confused.

Basically, as far as I am concerned, the game is rather broken when it comes to its treatment of special ammunition. That much is completely clear.

My experience over a rather large number of games is that the AI should quit worrying about wasting special ammo, since in 95% of the engagements the game represents, the vehicle in question has been dead long before it uses up ANY of its special ammo, much less all of it.

The game is overly liberal with allocating HVAP to American tanks to begin with, and then refuses to allow them to use it.

Fionns points about turning turrets at range, when he would engage and when he would not, are irrelevant. Whether it is tactically sound to engage at some range is irrelevant. The point is that given the order to engage, the AI should do so as efficiently and realistically as possible.

There seems to be a few issues happening here:

1. The AI does not know how to calculate its ability to get a kill with some given ammunition type. It thinks there is a chance for a kill when there often is not. This makes it think that it can use AP to kill a target when there is little or no chance of that happening.

2. The AI has no memory. Even after bouncing a dozen shots, it still thinks it can get a kill.

3. The AI thinks that its HVAP/APDS is much too precious. Having three rounds that you are unwilling to use is as useful as having zero rounds.

Anything else?

Any chance of getting some input from BTS on this issue?

Jeff Heidman

[Slapdragon]

With regards to tungsten, Fionns comments don't make sense in some respects and I agree with Jeff. Most tank and tank destroyers will dies with tungsten in their racks because they will not use it at all. Now -- if your routinely play Germans this is great! Because in all likelyhood it makes the American tanks more expensive without adding any value.

I have already stated this is a complex variable with likely 3-4 numbers interacting, and decoding these numbers will take a lot of work. At 500 yards E8 Shermans used tungsten around 1% of the time when their was a 15 to 1 round ratio. In a 440 yard free for all on open ground between 10 Panthers and 10 Shermans repeated five times the Sherman's with the same ratio never fired tungsten, and died with it still in their racks (7-10 kills ratio btw at 440 yards). At 700 yards the kill ratio was 4 to 10 and tungsten still was not used.

As for the proper kill range for a Sherman -- Fionn has misread my test. I tested for front turret kills at 500 meters and found them very rare, not overall kills at 500 meters -- hits on other locations that the turret are effective farther out. The thing about emprically testing means that when I test turret kills, that is all I can claim, and I never claimed that the Sherman could not kill at a farther range, only that front turret penetrations are rare, and that when the turrets of the Panthers turned they became much more vulnerable. The front turret may be Fionn's "ideal" hit location but it did not test out as ideal.

If on the other hand I was testing the best "kill range" of the E8 on the Panther, I would devise a new set of tests that did not use terrain to restrict me to front or side turret hits, use multiple range bands, and if I had enough time, test range versus aspect factors. My current findings though on this are just turret front kills, although my last set I relaxed that and just waited for a kill of anytype and saw what type of kill I got (since I was testing range kills rather than turret penetration)

Now that I am testing Hellcats it again seems the same -- a reluctance to use Tungsten in what boarders on suicide on tanks unless the Hellcat is seriously overmatched (KT v Hellcat) even though the Hellcat does not have a good chance of killing its target without them.

Now if I played Germans a lot, I would like to keep the Tungsten just the way it is, mostly unused. Each M10 that dies with its ammo hold full of tungsten after peppering me with AP is a good thing. All the other testers are saying is that tungsten is not used by the AI in situations it is clearly needed in, and that a Sherman historically fired it off if it had it, while in the game they die bravely rarely using it.

[This message has been edited by Slapdragon (edited 09-18-2000).]

[Juardis]

I'm not overly concerned with the use, or lack thereof, of special munitions simply because I don't purposely engage a Panther at long ranges. However, I'm not sure what I'm reading. On the surface, it appears that a Panther is the better choice vs. a Tiger or KT simply because the Amis won't fire tungsten at a Panther but will at a Tiger and thus stand a better chance of living.

But what good is a turret penetration test in real life applications (i.e., the game )? AFAIK, turrets are not purposely targeted (unless the target is hull down). So the overall chance for a kill may be listed as OK, but that means AP in the hull maybe, not necessarily AP in the turret. I don't care how my tank gets a kill, just so long as it does. If AP in the hull does the trick, so be it, save the tungsten. One of the tests above showed that fully half the kills came from lower hull penetrations. Combine that with the fact that almost all the American tanks have one or two rounds of tungsten when clearly they shouldn't lends credence to the fact that they're rarely used. (Yeah, I know, if you don't want them used then don't include them in the game.) But again, why use tungsten for a turret penetration whan an AP round can penetrate the hull? Unless the enemy tank is hull down where only the turret is seen, then using AP is fine by me so long as it does the job.

Is there something more that I'm missing?

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

Jeff Abbott

[Jeff Heidman]

I think we might be missing one big issue.

Do we *really* think that when some tank gunner sees a Panther in his sights he goes

"Gosh, that Panther looks like it is about 486.65m away. I know my AP has little chance of getting through that turret at this range, but, on the other hand, if I get a hull hit, I will almost certainly penetrate. But then, on the other hand, that turret is pretty dang tough. But gee, I was only able to steal 3 rounds of HVAP from the supply pool yesterday, hate to waste one on a hull hit on that Panther! Guess I will go with the AP..."

I don't think so. I think he says

"Holy ****!! A freaking Panther!!! Kill it! Kill it! I don't care if that is the last freaking HVAP round in Third Army, kill that thing NOW!!!"

Jeff Heidman

[This message has been edited by Jeff Heidman (edited 09-18-2000).]

[Jeff Heidman]

My point is that CM should err on the side of wasting tungsten rather than hoarding it. I would be willing to bet my next paycheck that a lot more tungsten rounds were wasted on targets that did not need them than AP rounds were used on targets that needed tungsten when tungsten was available.

Jeff Heidman

[Guest Mr. Johnson-<THC>-]

Have you guys all been using the same quality crews in your tests? Do regulars not know how to use T-rds proberly? Do elite crews know when its best to use T-rds? Back on 1.03 I remember I had a Stug-42 firing away on a Jumbo for like the last 5 turns of a game using only HE, the crew was regular I remember. The gun on the Jumbo was trashed and it was imbolized. My Stug had 5 rds of Case but never used them. Say if it was elite would it figure out at some point that HE was not going to punch through the armor and switch to case? I know this is a testing the 76 topic, but it has drifted into how does the Tac-Ai use Tungsten.

[Jeff Pattison]

I think this is precisely the kind of sanity check testing that should have been done, but was missed in the beta test program. I would be more than happy to help compile or verify results, if anyone wants to expand the testing.

All this talk about knocking Panthers out from the front reminds of a story I read a coupled years ago. An American Colonel in a tank brigade got wind that the Canadian Troop next door had discovered a way of knocking out Panthers from the front by ricocheting a shot off the ground in front of the tank into the belly armor. Naturally, he sent an LT to check the story out.

The LT arrived at the Canadian CP and after talking with the XO, he found that the story was true. The day before, a Churchill had indeed knocked out a Panther from the front by ricocheting a shot off the ground. When the LT asked if he could speak to the tank's gunner, the XO replied, "You will have go to hospital to talk to him. He's there recovering his nerve."

Jeff Pattison

[This message has been edited by Jeff Pattison (edited 09-18-2000).]

[Wolfe]

<BLOCKQUOTE>quote:</font><HR>How many tungsten rounds and AP did your Hellcats have?<HR></BLOCKQUOTE>

Four.

<BLOCKQUOTE>quote:</font><HR>Originally posted by PzKpfw 1 (quoting Fionn):

The AI evaluates 3 things when deciding whether to fire an AP or Tungsten

round.

1. Chance of hitting the enemy target.

2. Chance of an AP round penetrating said target.

3. Number of Tungsten rounds left.

So, if the chance of hitting the target is LOW then Tungsten won't be used unless there is literally no APCBC left.If an AP round has a good chance of hitting and killing the target then AP will be used and Tungsten saved.

<HR></BLOCKQUOTE>

Thanks for relaying the info, John. And to Fionn for the info.

But I still believe something is askew. I think the problem is the 76mm gun is seen as a real threat to killing a Panther frontally with AP (e.g. it get's an 'OK' rating for a kill). I ran some more tests last night.

I faced off 8 hulldown Panthers vs. 8 Hellcats armed only with AP. No Tungsten, no HE, no smoke, just AP. I ran the test through 10 times to try to get a bit closer to a statistically valid sample. I only counted turret hits, and ignored any upper hull hits to try to see what the odds were for an AP shell from an American 76 gun to penetrate the turret of a Panther G late. All tanks are Elite. Panthers are hull down and slightly lower than the Hellcats (I was trying to limit the number of upper hull hits; it worked to some degree). Distance is 500m.

* There were 471 hits on the Panther's turret that did no damage.

* 34 hits penetrated the Panther's turret.

* 5 hits penetrated at a weak point.

* 32 hits took out the gun, causing the crew to abandon the Panther.

My conclusion? The odds of an American 76mm gun firing AP rounds penetrating the Panther's turret at 500m is the same as the odds of getting a gun hit.

Incidentally, a couple of the turret penetrations were, indeed, from the first shot fired. But overall the odds of a kill are still quite low.

If you want to try yourself, here's the setup I used:

http://users.erols.com/chare/cm/Hellcat-vs-Panther_2.zip

It seems to me that with such low odds of success, the Panther should not show up as an 'OK' kill against a 76m American gun. I think if this was changed to either 'Low' or 'Rare', this would solve everyone's problems they have with the use of Tungsten rounds. After all, a tank who shows a 'Low' or 'Rare' chance of killing an enemy tank is much more likely to use Tungsten to get the job done (as I showed in my tests vs the KTs and others have experienced through gameplay).

- Chris

[RudeLover]

Forgive my ignorance here, but regarding the process of 'bracketing' with AP before firing HVAP / APDS, would that actually help a gunner?

Since the velocity of the projectile is so different, it seems to me that if you hit a fixed target with AP, then fired a round of APDS without adjusting anything, you'd miss on the high side, because the 'fall' of APDS would be significantly less over the same distance. Like the 'fall' of a fastball after several slow curves. Or was there a button that switched gunsights from AP to APDS aim? Or is the 'bracketing' process just to get gross distance information?

[Slapdragon]

We have not been testing crew quality (that I know of) but it is a good idea. Nor have we tested other units speciality rounds, like case, but that too is a good idea. The trick is that you have to test each variable seperately, and you have to do enough tests of the variable to create a statistical universe.

So, to test crew quality you make sure range is not a variable, and you see how often it gets used at a set range with a set unit facing the same tank. Then you change one variable, in this case crew quality, and you rerun the test. The minimum number needed to use parametic statistics is around 40 tests. At 40 tests you can have a largish standard error but it is good enough, at a hundred tests your standard error has fallen quite a bit and you can be more sure of your data (Panther front turret penetration was so low at 500 yards that it was eaten by standard error in my first test and I did not observe it, likewise I could have seen it twice as often. That is why multiple people doing the same test is good.

The important thing to remember is that this is a program, with a set of numbers ruling its behavior, and that makes it easy to crack as long as you isolate and test a single variable at a time. It is also important to remember that many events are incredibly unlikely but can happen, so be careful about war stories. My favorite is the King Tiger being killed by the Stuart -- sure it may be possible in some weird combination of factors but if we test it imperically we may find it happens so infrequently it is like a solar eclipse. So when someone comes on the list and says, the Stuart is to powerful and the KT needs beefing up, a Stuart killed my KT! Then you have to set up a test and see what is the actual chances.

Finally, if everyone uses the same test in the same environment The numbers can be compared with two tools, correlation and regression, to see if a certain factor contributes to a kill and how much it contributes. You also get a number called residual that tells you how much of what you are finding is caused by some other factor than you are looking at, called residual. Even if you leave the freshman statistics behind you can use grade school stats and just say: in 100 tests at 500 yards of a regular E8 crew shooting at a regular VG crew tungsten was used x times at 500 yards is good enough.

Front turret penetration is important for tank duels because it tells you the range that a tank is likely to be able to tackle its opponents in a gunfight: ie. when two tanks spot each other at suddenly and start shooting. This is especially true with defending units since they can more often choose a hull down attitude than an attacker who has to move from place to place.

[This message has been edited by Slapdragon (edited 09-18-2000).]

[Slapdragon]

For anyone who needs a test range, I have posted my firing range at: http://www.slapdragon.org/combat/firingrange.txt . Some PC format computers may have trouble downloading the file but I do not use Zip files on this server, only raw data files, so you may need to do a right click download to get it. It is listed as a .txt so you must manually change it to .cmb -- this keep me from having to add a new MIME type to my computer.

[This message has been edited by Slapdragon (edited 09-18-2000).]

[This message has been edited by Slapdragon (edited 09-18-2000).]

[Slapdragon]

One further note:

This is not a critique of the engine, since this is exactly the place were a small design team has trouble : ironing out all of the little questions, because it takes a lot of drudge time to properly document a problem such as tungsten. The beauty of open source software (which this not, I know, but this discussion is similar to an open source discussion) is that lots of people look emprically at problems and come to a concensus solution. It can be very messy, since you find many people have a vested interest in the weirdest status quo.

A very similar thing has happened in other games. Steel Panthers on the alt.games discussion group had a discussion of play balancing in Desert Scenarios between the British and Germans. Turned out that the engine never really could handle handing out points so that the game was fair, and still hand out points so that fighting in mountains and towns was fair. So tourney strategy in Steel Panthers was: choose the Germans. The reason turned out to be the 88. It would get wiped out, but would kill enough Brits that it was worth like 8 times its printed value. It took lots of arguments and back and forth to get that established, and lots of people who relied on the 88 to tip the game to a hands down got upset. (Of course SSI never fixed it that I know -- we players just hacked work arounds and kept going.)

Steve Jackson

[Vanir]

I'm glad to see Fionn had chipped in his two bits on this one. That having been said I join Jeff and Slapdragon in scratching my head...

<BLOCKQUOTE>quote:</font><HR>: The AI evaluates 3 things when deciding whether to fire an AP or Tungsten

round.

1. Chance of hitting the enemy target.

2. Chance of an AP round penetrating said target.

3. Number of Tungsten rounds left.<HR></BLOCKQUOTE>

This pretty much confirms what we already suspected, but it's nice to hear it from someone who is more familiar with the game mechanics.

<BLOCKQUOTE>quote:</font><HR>

So, if the chance of hitting the target is LOW then Tungsten won't be used unless there is literally no APCBC left.If an AP round has a good chance of hitting and killing the target then AP will be used and Tungsten saved. <HR></BLOCKQUOTE>

Is an 82% hit percentage low? The TacAI sure seems to think so. (I understand you may have sent your email to John after I reported this tid-bit in one of my follow-up posts). After one round of firing (6 or 7 shots) the 76 has an 82% percent chance to hit the Panther at 700m. But it still would not switch to tungsten.

<BLOCKQUOTE>quote:</font><HR>As for the "Tungsten isn't supposed to be used as a first round" thing. It ISN'T IF the tank has a normal loadout. A normal loadout is one in which Tungsten is "rare" and so the AI uses AP shells to bracket the target.

OTOH if you skew the normal loadout and add more Tungsten then the AI says, "Hell, I have lots of Tungsten and this is a big enemy tank... I have enough Tungsten to waste 1 or 2 on bracketing shots ( in the hope that one of the braketing shots will hit and kill the tank." <HR></BLOCKQUOTE>

Your basicly echoing what M Hofbauer said earlier so my response will be the same: There is no evidence that tanks ever bracket with AP and then switch to tungsten. I understand that this is how its supposed to work, but both Slapdragon and I have been unable to produce this course of action in our tests using a normal loadout. If you reread my test you will see that the Jumbos fired 5, 8, and 9 rounds respectively before dying with all tungsten still in the ammo holders. How many shots are required before the target is bracketed? More than 9 it seems. In every test I've ran, when the tanks start shooting with AP they keep right on shooting AP until they die never switching to tungsten.

<BLOCKQUOTE>quote:</font><HR>

Again, you can't simply take a statement which is obviously made when referring to normal loadouts and then apply it to an abnormal loadout and then wonder why it no longer applies.<HR></BLOCKQUOTE>

I simply don't understand why you make this statement. We have done no such thing. In my first test the loadout I used was 30 AP, 30 HE, 3 tungsten, 5 smoke. This is hardly an abnormal loadout. In fact, I purposely choose that loadout because it was typical. The whole point of all this is that the only way we can ever get Allied tanks to use tungsten is to give them abnormal loadouts.

I will say it one last time: with a loadout of 30 AP, 30 HE, 3 t and 5 smoke the Jumbo 76s never used tungsten; never bracketed and then switched, even after firing up to 9 rounds and achieving a to-hit chance of at least 82%, even after watching a total of 18 previous hits bounce harmlessly off their targets. If anyone can expain to me why this is exactly how it was intended I want to hear it. Explanations of how its supposed to work aren't cutting it, because the impirical test data (not just mine) proves that it isn't working the way its supposed to.

I hope I don't sound cross or angry, that's not my intention. I'm simply frustrated that I can't seem to get what seems to me to be a clear as day point across to people that I know to be quite intelligent. It makes me wonder if my communication skills are not as good as I thought

<BLOCKQUOTE>quote:</font><HR>Jeff:

1. The AI does not know how to calculate its ability to get a kill with some given ammunition type. It thinks there is a chance for a kill when there often is not. This makes it think that it can use AP to kill a target when there is little or no chance of that happening.

2. The AI has no memory. Even after bouncing a dozen shots, it still thinks it can get a kill.

3. The AI thinks that its HVAP/APDS is much too precious. Having three rounds that you are unwilling to use is as useful as having zero rounds.<HR></BLOCKQUOTE>

Well, I'm glad to see someone thinks we're making sense here I think you've summed up the problems quite well (better than I have it seems). I personaly think #1 on your list is the big culprit here. At 700m the chance for the 76 to kill a Panther frontally is given as "ok". I think we all know this is wildly optimistic. Even Fionn says he considers 500m to be about the cutoff point. If the TacAI is grossly over estimating its chances for a kill (is there any doubt?) this could easily skew the internal equation the game uses to determine if tungsten is appropriate.

P.S.: I agree with the earlier statement that if anything, tanks should be a little too eager to use tungsten rather than too reluctant. This would be especially true of less experienced crews who would be more likely to panic and grab the Magic Bullets right away. More experienced crews would be more likely to posses the disipline to bracket when bracketing isn't likely to get them killed.

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

No, there will be no sequels. Charles and Steve have given up wargame design in disgust and have gone off to Jamaica to invest their new-found wealth in the drug trade. -Michael emrys

[This message has been edited by Vanir (edited 09-18-2000).]

[Slapdragon]

To echo Vanir here I have been tweeking ammo loads outs in increments to find out when the AI feels "comfortable" with using tungsten rounds (silly term to consider a couple hundred lines of code "comfortable"). It was not until I got the tungsten to 2:1 ration (10 tungsten and 20 AP) that I could even get the AI to use it in the E8. The M18 is turning out the same except it uses them faster because it carries less ammo (higher ratio) while my cross check on the Jumbo 76 is the same.

[Simon Fox]

This is an interesting topic and some good points have been made all round. Fionn's points are relevant, especially the point about accuracy but I still think tungsten usage seems awfully low. Maybe the AI is just being a little too sensible

Firstly, US (and German) subcalibre rounds were inferior to the British APDS which was less prone to shatter and also IIRC more accurate. Therefore British units may be more likely to use their APDS (which was more plentiful anyway).

Secondly, HVAP and APDS are less accurate than conventional rounds. As RudeLover so rightly pointed out the flight/trajectory is quite different so 'getting the range' with APCBC is actually quite dumb at longer ranges. At least some of the alleged inaccuracy of APDS was due to the practice of bracketing with APCBC. British firing tests clearly demonstrated this. When APDS was used for bracketing the accuracy substantially improved. Not to the level of APCBC but out to a certain range it was very acceptable. Therefore British practice recommended that:

(A)Identify target and range

(B)Use appropriate ammo from the 1st round

WO 291/762, "Accuracy of APCBC/HV shot and AP/DS when fired from the 6-Pounder Gun mounted in the Churchill IV."

[Jeff Heidman]

ping.

Does anyone know if Steve or Charles are aware of this discussion? I would be interested in their perspective on the issue, as would mst of us, I think.

Jeff