Tank gun test, extreme ranges. (Warning: large pics inside)

[Andreas]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Jeff Duquette:

The following is a partial list of reference material utilized by Janusz Piekalkiewicz in his book “The German 88 Gun In Combat, The Scourage of Allied Armor”, Schiffer 1992. Could you translate what the tittles of these manuals are. Also what is your thinking as to what these references are. Thanks much.<HR></BLOCKQUOTE>

I am not Marcus, but here goes anyway. They are as far as I understand it the technical manual and SOP documents.

<BLOCKQUOTE>quote:</font><HR>Originally posted by Jeff Duquette:

"Der Reichsnninister der Luftfahrtund Oberbefehlshaber der

Luftwaffe, Einsatz, Verwendung und Kapoffuhrung der Flak-

artillerie", LDv 400/10, Teil III, Berlin, September 1942<HR></BLOCKQUOTE>

The Imperial Secretary of State for Air Traffic and Commander of the Air Force. Combat, use and combat command of the AA artillery. Air Force SOP 400/10, Part III, Berlin, September 1942.

<BLOCKQUOTE>quote:</font><HR>Originally posted by Jeff Duquette:

"Der Reichsminister der Luftfahrt und Oberbefehlshaber der

Luftwaffe, Ausbildungsvorschrift fur die Flakartillerie", LDv

400/10, Teil V, March 1943<HR></BLOCKQUOTE>

The Imperial Secretary of State for Air Traffic and Commander of the Air Force. Training manual for the AA artillery. Air Force SOP 400/10, Part V, Berlin, September 1942.

<BLOCKQUOTE>quote:</font><HR>Originally posted by Jeff Duquette:

"Der Reichsminister der Luftfahrt und Oberbefehlshaber der Luftwaffe, Ausbildungsvorschrift fur die Flakartillerie", LDv 400/10, Teil IV, April 1943. <HR></BLOCKQUOTE>

The Imperial Secretary of State for Air Traffic and Commander of the Air Force. Training manual for the AA artillery. Air Force SOP 400/10, Part III, Berlin, September 1942.

Piekalkiewicz also uses memoirs, KTBs, and articles from scientific journals as sources.

Apologies if there are inconsistencies in the translation. It should be pretty accurate, but some of the nuances might have escaped me.

BTW - the English title of the book is just corny.

[M Hofbauer]

edit: I see that Germanboy already did it.

[ 10-03-2001: Message edited by: M Hofbauer ]

[Andreas]

<BLOCKQUOTE>quote:</font><HR>Originally posted by M Hofbauer:

edit: I see that Germanboy already did it.

[ 10-03-2001: Message edited by: M Hofbauer ]<HR></BLOCKQUOTE>

Yeah, but I can not give your opinion on their usefulness, Du Nase

[Jeff Duquette]

Thanks Germanboy for the translations. Also thanks Marcus as it looks like you were about to post the translations.

[Holien]

Fair point Maastrictian...

I do not see why range makes a difference to Jasons argument.

BTW I don't disagree with Jasons point, I just think there is an element of wasted ammo on on targets and it should be factored in.

Not dismissed.

Back to the thread...

H

P.s This continual denouncement of Finns is not healthy. They have enough to put up with let alone being ragged at every opportunity on this board.

[tss]

I thought that I should elaborate my comment about the number of unknown variables. I'll do it by an example:

Let's start by designing a mathematical formula for AP ammo usage, so that things will look neat and there's an illusion of respectability:

R = N * D * T * S * C * W / K

Where

- R = Ratio of rounds used per killed tank.

- N = Number of AT rounds produced

- D = Proportion of rounds left in depots

- S = Proportion of rounds lost in supply chain

- T = Proportion of rounds used in training

- C = Proportion of rounds captured by enemy or destroyed to prevent capture

- W = Proportion of rounds wasted by firing them at non-vehicular or already KO targets

- K = Number of tanks knocked out.

[To be exact, all those terms corresponding to the proportions should be of the form (1 - X) but it looks nicer the way it is above, I'll just turn the proportions the other way round.]

Let's use Jason's figure of 400 produced rounds per AT weapon and a wild guess that each gun destroyed one enemy AFV to simplify the calculation (we can then throw K out and use N=400). We have now to assign some concrete figures to remaining five parameters.

First, consider D. Germans were certainly quite efficient, so a guess that 90% of the shells left the depots sounds reasonable and we can set D = 0.9. Similarily, it sounds reasonable to say that at most one shell in twenty was lost on way to front, so S = 0.95.

Though training was important, the live ammo usage there was much smaller than at front. Perhaps 5% is quite close, so T = 0.95.

Because of German setbacks of 1944 and -45 the number of rounds lost to enemy was quite big, at least one third. So, let's use C = 0.67.

Now we have to estimate the last free variable. Again, it is safe to assume that about 1/3 of ammo was wasted on invalid targets and W = 0.67.

Plugging these figures in we get:

R = 400 * 0.9 * 0.95 * 0.95 * 0.67 * 0.67 = 145.8

Now we can conclude that a gun fired on average 146 rounds at an enemy tank for a kill so CM's guns are way too accurate.

On second thoughts, that seems to be a little excessive so let's adjust the parameters slightly.

Certainly, by late 1944 German transportation system was having serious troubles and since German production increased in the latter part of the war, we can estimate that perhaps as high proportion as 1/3 total AP ammo production never left the main depots and factories, so D = 0.67.

Also, thinking about it, partisans certainly inflicted a heavy toll on German supply trains and Allied aircraft interdiction sent many supply trucks to heaven, so maybe as much as 1/2 of rounds got lost on way and S = 0.5.

Germans trained their troops pretty well, so guessing that T = 0.85 doesn't sound too unreasonable to me.

In 1944 Germans had to blow up lots of their great supply centers because of Allied advances on all fronts and a number were captured intact. It may be that as much as 60% of remaining ammo was lost this way, and C = 0.4.

Finally, lots and lots of rounds were wasted by shooting at supposed targets because gunners thought that better safe than sorry. It is not impossible that 95% of rounds were fired this way, and W = 0.05.

Using these improved figures we get:

R = 400 * 0.67 * 0.5 * 0.85 * 0.4 * 0.05 = 2.3

So, we can conclude now that on average each tank kill required only 2.3 shots.

At this point one might start to question the validity of these two results. After all, changes in the parameters caused a difference of two orders of magnitude to the result. And all figures that were used could be justified by some argument (though in most case by quite weak argument). I don't know the correct values of any of the seven parameters. I strongly suspect that no-one knows the correct value of any of the parameters (except possibly N and T, if records survived the war).

Now, what is the moral of this post? By choosing the parameters in a suitable way, you can get get out any value you want for R.

This is why I think that the whole reasoning process is flawed.

- Tommi

[Slapdragon]

The ragging on Finns is not, actually, all that fair. Every group has a uber movement who cannot see the forest for the trees. Lots of Finns on this board have never showed themselves to be unusually nationalistic. I know Ari and Tss are very open minded.

The Uber Finn thing comes from some silliness on the part of two men (Hill and Greenwood) and from some narrow minded commentary by some Finns quite some time back.

To prove the uber hypothesis (that there are just not uber Finns) look at the recent explosions by a small group of Australians, another set of explosions much earlier by a larger set of Germanophiles (who were mostly not German), and a more recent out break of uber Americana.

Since every country is equally screwed in CM there is no point in blanket statements against one group of people. In fact, the only evidence I have seen presented was an analysis of 500 rugged defense games that found Germans won more games, and even that is open to debate.

As for Jason C, his idea is a good one, and possibly valid, just with execution flaws. The reason is because of a basic statistical and quantitative reasoning error that you could call the error of potential.

Germany manufactures x million 88mm shells. German 88s are credited with y hundred hard kills on enemy tanks. If the variables fit perfectly together then you would expect that x/y=# shells fired per kill.

It does not work that way because there are several attrition variables (which is not the same as statistical attrition) at work. First is that not all shells are expended in battle, and then not all shells are expended at the intended target tanks. Then there is the loss of shells to logistic disruption.

So the situation is similar to what faces a marine scientist who measures phytoplankton to determine bio productivity in an ecosystem. You find that there in 36 tonnes of phytoplankton per km2 of ocean and expect to find that there is 3.6 tonnes of Krill and primary consumers, .36 tonnes of secondary consumers, and .036 tonnes of tertiary consumers. However, you cannot make the claim that finding 36 tonnes of phytoplanton = a habitat support rate of 36 kg of tertiary predator. The reason is that all of that energy influx into the system may never reach the higher levels for hundreds of reasons.

His numbers though would be much better if he could get to the number of rounds issued per gun over its lifetime (and you get this by looking at unit records). Not perfect, but at least you know the ammo was at the front.

[Wreck]

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

Now, what is the moral of this post? By choosing the parameters in a suitable way, you can get get out any value you want for R.

This is why I think that the whole reasoning process is flawed.

<HR></BLOCKQUOTE>

It is true that by choosing the parameters suitably, you can get out any result you want. For instance, if I take your latter set of numbers, but use the assumption that actually more like 80 shells were fired at bushes and barns per one fired at an enemy tank (rather than your rather generous assessment of 1:20), then we find out that every shell fired at an enemy tank by the German 88s actually killed two enemy tanks!

Seriously. Yes, you can get any result, but only by picking numbers that may well be absurd, and which can be criticized. In your case, even by picking rather extreme numbers in your second example, you still end up with 2.3 shots fired per kill. Well, that is still an average that lets us know something. And it is a worst-case lower bound, and in fact, not credible.

As for your formula, well, consider the following way of looking at the terms. Most of the factors are ways that shells might not get to the individual tank or gun at the front that is supposed to fire them. These are D, S, T, and C. We can therefore multiply them together to find the proportion of shells produced that reached a weapon at the front:

D * S * T * C = 0.1139

So, of the 400 shells produced per gun, your assumptions are equivalent to the assertion that only 45 shells reached each weapon.

But 45 shells, is about the average taken into a single combat by a Tiger or 88mm AT. (A bit more, IIRC, but close). So we can show, even without knowing precisely which one of the variables is set wrong, that in toto they are wrong.

This is how using numbers is, in fact, clarifying and helpful. And in your case, even by stretching the bounds of credibility in your assumptions you still came out with a kill-ratio of 2.3 kills per AT round fired at a tank... obviously the average for long ranged fire, being less accurate than short, will be worse.

[rexford]

Recon by fire is a common tactic where enemy may be hiding and you want to rattle their cage and obtain some motion or return fire. The initial use of ammo for recon by fire does not count towards accuracy since it not aimed fired for K-O purposes, per se.

Tanks that are destroyed or knocked out and lost also cost ammunition.

At Kasserine Pass, U.S. troops held their rifles over the trenches and fired wildly hoping to do something without coming into harms way (story from a local who was there or close to Nebelwerfer hits). It is possible that a few tanks did most of the killing and the rest stood around and fired off a few rounds to seem as if they in the thick of things.

My favorite movie, The Unforgiven with The Man, El Hombre, Clint Eastwood, has a good explanation of how one man outshoots a group. This explanation may also apply to tank warfare.

As Clint explained, one figures out who is the most likely to shoot straight and you get them first. The others will be so scared once the shooting starts they will lift their guns and fire away without aiming. Those you pick off later on after the main threats are done away with.

If one is lucky they can get the straight shooters without being hit, and then pick off the scare-dee-cats at leisure.

Oh, by the way, how many rounds were spent on practice fire or used to align gun and sights. How many 88 rounds went to the bottom of the Med Sea after a British ship or plane hit a cargo run bound for North Africa.

In Barkmann's Corner, the Panther continues to put rounds in the knocked out Shermans after they are disabled so they will burn.

[tss]

Wreck wrote:

In your case, even by picking rather extreme numbers in your second example, you still end up with 2.3 shots fired per kill.

I think that my first example is actually more relevant than that one since the numbers there are individually more plausible. (But I want to emphasize that any connection between them and reality is purely coincidental). I could accept each of them individually, but together they give a ridiculously poor shot/kill ratio.

I personally have no trouble with the claim that in long range battles AT guns had to use on average 6-12, or even more, rounds per kill. I wouldn't be too surprised if it turned out to be closer to 20. I just don't think that you can really justify accuracy claims based on ammo production, because there are too many unknown holes there.

And it is a worst-case lower bound, and in fact, not credible.

Like I mentioned above, in this case the upper bound causes a more serious problem since the figures in it all seem credible at the first sight so it isn't even really the least upper bound.

So, of the 400 shells produced per gun, your assumptions are equivalent to the assertion that only 45 shells reached each weapon.

Only if we assume that all guns reached the front. And that all guns had time to fire even once their ammo alotment. However, I don't want to try to defend that estimate since I just wanted to throw in some plausible figures (where "plausible" means here "not completely impossible").

So we can show, even without knowing precisely which one of the variables is set wrong, that in toto they are wrong.

Yup. But you can still easily get differences of one order of magnitude using only plausible values.

obviously the average for long ranged fire, being less accurate than short, will be worse.

Obviously.

- Tommi

[JonS]

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

... Oh, by the way, how many rounds were ... used to align gun and sights...<HR></BLOCKQUOTE>

Actually, in this case the figure is zero. Oddly enough, one doesn't need to fire to ensure that the barrel is aligned with the sight, instead a gunners qudrant is used.

Well, that is the case with current field artillery, and I suspect it hasn't changed all that much over the intervening period. But then again, what would I know, eh Slap?

[tss]

rexford wrote:

Tanks that are destroyed or knocked out and lost also cost ammunition.

I knew that I had forgotten some important factor.

It is possible that a few tanks did most of the killing and the rest stood around and fired off a few rounds to seem as if they in the thick of things.

That seems plausible.

Risking to rise Jason's ire, I just checked that 5 highest-scoring Finnish StuG gunners got little less than 50% (42/87) of credited tank kills. Since there were approximately 30 StuGs that participated in combat (I haven't compiled a full list, yet), it means that roughly 16% of vehicles got 50% of kills.

Once again, this situation is not likely to be very good example for the general case. The active combat period was quite short, only a little over month and there wasn't time for large performance gaps to develop. In total, 19 gunners got a Tank-Destroyer's Badge, with the kills ranging from 1 to 11.

- Tommi

[PJungnitsch]

In air combat that type of thing is called the 'ace factor'. During WWII it was estimated that the top 5% of pilots got 80% of the kills, or some such. Pilots were divided fairly sharply into either 'eagles' or 'turkeys', with very little inbetween.

[Vanir Ausf B]

IIRC it was 11% of the pilots got 90% of the kills.

[ 10-03-2001: Message edited by: Vanir Ausf B ]

[Jeff Duquette]

Now here is a fellow that obviously took the time to actually read the TigerFibel and knew that you can miss with the 88 out to ranges of 1200m. From: Hans-Joachim Jung's “Panzer Regiment Grossdeutschland”(Page 158).

<BLOCKQUOTE>quote:</font><HR>The actions of Feldwbel Rampel of the 11. Kompanie... He was supposed to take his tank to the maintenance facility on account of transmission damage. Since he was still mobile, he procured ammunition and a loader from another disabled Tiger and then received more ammunition from a one-ton prime mover. Rampel then engaged forty attacking enemy tanks after an 88mm Flak gun positioned there withdrew. He almost knocked out a friendly Panther heading back to the facility. At a range of 1200 to 1500 meters he opened fire on the 40 attacking T-34's and KV 1’s, and every shot was a direct hit. Rampel destroyed seventeen tanks, causing the enemy to break off the attack. With his Tiger hit in the turret and main gun, Rampel drove back to the maintenance shop and transferred his wounded crew members to the troop aid station.<HR></BLOCKQUOTE>

[Brian Rock]

<BLOCKQUOTE>quote:</font><HR>At a range of 1200 to 1500 meters he opened fire on the 40 attacking T-34's and KV 1’s, and every shot was a direct hit. Rampel destroyed seventeen tanks, causing the enemy to break off the attack.<HR></BLOCKQUOTE>

This must be the most inept armor attack in history. If this story is taken at face value an entire tank battalion (+) was unable to advance more than 300 meters against a single tank.

Surely someone somewhere has made a mistake and the Germans really did win WWII.

[PJungnitsch]

<BLOCKQUOTE>quote:</font><HR>This must be the most inept armor attack in history. If this story is taken at face value an entire tank battalion (+) was unable to advance more than 300 meters against a single tank.<HR></BLOCKQUOTE>

There's a reason why the Tiger has a more fearsome reputation today than say, a Valentine or Crusader

And AFAIK most of that reputation was built when they could outrange their enemy:

"...two Tigers acting as an armoured point...Normally the Russians would stand in ambush at the safe distance of 1235 meters...the Tigers...made use of their longer range...within a short time they had knocked out 16 T-34's which were sitting in open terrain, and when the others turned about, the Tigers pursued the fleeing Russians and destroyed 18 more tanks."

[JasonC]

One fellow mentioned an analogy from ecology, but the actual, better parallel is life tables in interactive population studies. Which I know a little about, and are the sort of science parallel behind the use of production numbers to test hypotheses about probabilities for some sort interaction.

As in, two dobuggers create 100 little dobuggers per generation, population remains constant, therefore kill factors multiplying out to .98 are to be allocated among the causes of mortality, with so-n-so many larvae, so-n-so many live pupae, etc, around to new dobuggers again. It is a common enough form of reasoning.

One fellow wanted to pretend it is entirely arbitrary by giving two sets of numbers he doesn't believe himself (which is usually known as lying, but here passes for rhetorical argument). But he doesn't seem to notice how little they support that conclusion. When you have to assume strictly unbelievable things to make a conclusion consistent with the model, the conclusion is strictly unbelievable. His numbers actually exclude the intermediate hypotheses he proposed.

He also used redundant categories, probably without realizing it. Over the whole war, fired, destroyed, and overrun, together, are exhaustive. Something in the supply chain is later either fired or destroyed or overrun. It is not like the war is still going on.

But most revealing was the last part of the 2.3 forced conclusion, where he needed 19 out of 20 rounds fired at anything but their intended targets in order to justify the ending number he wanted for rhetorical purposes. He doesn't seem to notice the practical conclusion that would follow from this, however.

The reason this thread is here is because somebody thought tanks should be killing enemy tanks with fewer shells fired. But if the "spray the barns" idea is believed, then that is the last thing those who believe it should call for. They would want a Tac AI that fires 19 out of 20 rounds at phantom targets, and only rarely even points in the general direction of enemy tanks at all. This is not exactly what the original poster wanted. Its tactical effect would be the same as - or even a more extreme version of - many shots being fired and missing. Surely it makes no tactical difference to him whether the Tac AI fires 35 rounds at empty space and hits one tank with the next 5, or fires 40 to hit one tank.

What that argument cannot justify is all AFVs coming onto the map with full ammo loads, and holding their fire until enemy AFVs present themselves, and then dispatching them with 2 rounds apiece. If all the rounds were historically fired at misidentified targets, then surely misidentified targets belong in CM. Where lo, they will perform the exact same tactical role as misses, because misses by any other name is what they are. Or are tanks to develop the tightest supposedly ahistorical fire discipline magically, because a human has assumed command?

No one has yet told me any reason a tanker should not have fired at ranges with 3-5% hit probability. Since hit probability undoubtedly declines with range, presumable - at least on steppe or desert I suppose - there is some range at which that is the hit probability for a given gun. Why not fire at that range? - if the target would still be penetrable, as it should be for the big guns most under discussion when we talk about long range shooting.

However long you think that distance might be, I'd still like to hear anyone's argument against shooting at that range. Because those maintaining that of course nobody would fire unless the hit probability was 20%, is necessarily maintaining that the tankers held fire through the whole range of hit probabilities from 3-5% up to his suggested, improved level of accuracy, whatever that is. If anyone honestly believes that is the case, they must have some obvious reason why they think firing still farther wouldn't be attempted. Share it with us.

Some are concerned about the life-table style of analysis, because they doubt it can achieve the precision necessary to pin down actually average accuracy figures. In practice, it can -falsify- some claims made about average accuracies, by providing bounds, much more readily than it can establish real average figures within any narrow range. But it could easily do even the latter in principle, if data were collected on other "lines" of the ammo "life table". The interrelations of numbers check each other, and it is essentially impossible to maintain grossly false hypotheses about complicated historical processes, if any real data to speak of is accessible to test the necessary intermediary claims.

And there are some such checks available, even without reducing them to numbers, just from common sense. If 9/10 shells made never reach the front, all a side has to do to reap large economies is stop making the excess shells that only stuff rear area depots. Assumptions about the behavior of the participants that reduce them to enourmous boneheads are inherently unbelievable.

And other cases cross check things like distribution loss claims. For instance, if you apply the loss claims for channels to small arms, would the Germans only be able to oufit 5 divisions with rifles over the entire war? If so, the loss claims from channels are not believable. Marginal differences are possible across categories of arms, but not orders of magnitude.

One fellow mentioned the loss of ammo when AT weapons are destroyed, but apparently overlooked the fact that I had already mentioned it. It won't account for much of the discrepency, only around 1/10th. Because the rounds per weapon in the field at a given time is significantly less than the rounds made per weapon. And because a lost weapon could have any ammo state from full to out at the time of its loss, so the average ammo loss along with the weapon is likely to be on the order of 1/2 its typical field load, or only marginally higher.

Another fellow though that perhaps engagements without result are a seperate category, but they are not. They are simply misses. If a tank fires 3 times and misses in one engagement, and 5 times in another with one kill, and 4 times in another missing, and 4 times in another with a second kill, then it has fired 16 times and killed 2, with a hit probability on the order of 12.5%. It is not like its hit chance is "really" 20-25%, from just the two "above average" occasions when it got kills. Yes, all 16 rounds fired count, of course. But the misses in engagement #1 are not in a difference category from the misses in engagement #2. Both contribute to the denominator that lowers average accuracy.

There was also some discussion of the ace factor, which is an interesting subject, certainly. But people often address it with considerable naivety, statistically speaking. You expect a signficant amount of "acedom" without any actual ace effect causing it. Just from randomness. Actual ace effect is a different thing, and it takes more to spot it than noting x% of the shooters got y% of the kills.

The mathematical entity involved is called a negative binomial distribution. If every crew has the same odds of winning a given engagement, you will still get a some with "runs" much longer than others, before a defeat. And when the two things being measured aren't exactly the same, it can get still more complicated. This is common when one is tracking how many -vehicles- are taken out by one -crew-, when the crew may have several vehicles of its own in sequence. That is, the probability of crew loss is not one minus the probability of enemy vehicle loss, because there are also own vehicle losses in which the crew survives.

As an example of how these things work, you can ask the question, what would the odds of crew loss per trial of tank vs. tank have to be, for an expected 1 crew out of 1350 to run up 130 vehicle successes? The answer is about 15 to 1 against. Which doesn't mean the crew in question KOs the enemy vehicle 15 times for every loss of its own vehicle. It means it is 15 times more likely in one trial to KO an enemy vehicle as to lose not only its vehicle, but the crew as well.

To find an "ace" effect, you have to first see how much "acedom" you'd expect even with uniform crews, each with some probability of winning their next engagement rather than losing it. It will not be nothing, especially in a sizable population and in cases where the absolute probability of success per trial, for the whole population, is high. (It is easier to get long "runs" with higher chances each, and with more having a chance).

An actual ace effect will show up as a -skew- in such distributions, away from the "acedom" due merely to randomness.

For what it is worth...

[JasonC]

On the uber Tiger tales, some people don't seem to be able to detect irony I guess. Meanwhile, back in reality, the Russians went over the field at Kursk and measured the holes in their tanks. 20% were from 88mm, including towed guns, the Elephants, and Nashorns, as well as the Tigers. Most of the holes were from 75mm.

[tss]

First, I would like to apologize Jason for the overly insulting tone of my yesterday's post. I should have let my head cool before posting.

JasonC wrote:

(which is usually known as lying, but here passes for rhetorical argument).

I see that we have a different definition for lying, then.

When you have to assume strictly unbelievable things to make a conclusion consistent with the model, the conclusion is strictly unbelievable.

Umm... My conclusion was that by choosing the unknown parameters suitably you can get any figure between 2-150 to the number of rounds / kill.

His numbers actually exclude the intermediate hypotheses he proposed.

What intermediate hypotheses? I wasn't aware I had any.

He also used redundant categories, probably without realizing it.

While realizing it. I just thought it to be a good idea to explicitly show the kinds of things that are there.

But most revealing was the last part of the 2.3 forced conclusion, where he needed 19 out of 20 rounds fired at anything but their intended targets in order to justify the ending number he wanted for rhetorical purposes.

I suppose that you have some concrete figure based on first-hand sources to substitute there, then.

I think I made it quite clear that any connections between numbers in my post and reality were strictly coincidental. Though, I have to admit, that I think the same holds for your figures.

Surely it makes no tactical difference to him whether the Tac AI fires 35 rounds at empty space and hits one tank with the next 5, or fires 40 to hit one tank.

If those two cases don't make any difference to you, then I have to conclude we have to agree to disagree here as continuing this discussion would be pointless to the extreme.

Note that I'm not demanding that CM tanks should fire at non-existing targets. While being arguably more realistic, it would be a mistake from playability viewpoint.

However, arguing that those rounds that would have been wasted in a real combat should be factored in by lowering general accuracy, would be in my opinion quite strange.

Where lo, they will perform the exact same tactical role as misses, because misses by any other name is what they are.

Now, let me ask this one straight question:

Do you think that CM should artificially lower the accuracy of targeted shots so that the average ratio of shots/kill is close to whole-war average?

My viewpoint is that in tactical combat it is more important to get the accuracy of correctly targeted shots correct. Your mileage may vary.

Or are tanks to develop the tightest supposedly ahistorical fire discipline magically, because a human has assumed command?

Given that the game currently has extremely unrealistic modeling of chain of command, I don't see any trouble with unrealistic fire discipline. I have more trouble with adhering to some unknown average values.

No one has yet told me any reason a tanker should not have fired at ranges with 3-5% hit probability.

Here's two:

1) You may reveal the position of your main line of resistance too soon so that the enemy has time to call artillery and air attacks. This is not so serious with tanks, but with AT guns too early commitment just causes unnecessary deaths.

2) If you wait, the enemy may get closer and you have better changes inflicting heavy casualties before they have time to react.

Now, before these get turned into strawmen, I'd like to explicitly point out that after the combat has begun there's much less reasons to hold low-probability shots.

In practice, it can -falsify- some claims made about average accuracies, by providing bounds, much more readily than it can establish real average figures within any narrow range.

But in this case you can get an upper bound of ~150 shots/kill with figures that are still plausible. Should we expect that in CM a 6-vehicle StuG battery that fires all its AP ammo gets on average only one effective hit?

You expect a signficant amount of "acedom" without any actual ace effect causing it. Just from randomness.

True, and a good point.

- Tommi

[M Hofbauer]

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

On the uber Tiger tales, some people don't seem to be able to detect irony I guess. Meanwhile, back in reality, the Russians went over the field at Kursk and measured the holes in their tanks. 20% were from 88mm, including towed guns, the Elephants, and Nashorns, as well as the Tigers. Most of the holes were from 75mm.<HR></BLOCKQUOTE>

I don't see your reasoning.

These two statements do not have to contradict each other at all. In fact, they do *not* contradict each other since we *know*that the majority of guns was 7.5cm, not 8.8cm.

[c3k]

JasonC,

Your frustration with the resistance to your numerical analysis is showing. Posting (or insinuating) that tss is "lying" is not constructive and belittles your contribution.

Post Kursk analysis showing 20% of Soviet tanks KO'd to 8.8cm rounds, the rest to 7.5cm rounds? If so, what is the ratio of 8.8cm guns to 7.5cm guns? Ammo used and range engaged? This was an exceptional attack on an exceptional defense. Given that fact, I'd expect the range of engagement to be significantly less on average than typically expected.

Further, looking at ammo production numbers, what percentage of 8.8cm ammo went to anti-aircraft guns? What % was actual armor piercing? At the end of the war, how many shells were captured? None? Or a few million?

Statistical, or numerical, analysis can be a useful tool. As tss showed it can help to broadly define boundaries. As Secretary of Defense McNamara showed, it is useless as any sort of measure of efficiency or success.

[Slapdragon]

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

Actually, in this case the figure is zero. Oddly enough, one doesn't need to fire to ensure that the barrel is aligned with the sight, instead a gunners qudrant is used.

Well, that is the case with current field artillery, and I suspect it hasn't changed all that much over the intervening period. But then again, what would I know, eh Slap? <HR></BLOCKQUOTE>

Rexford may be referring to rounds used to "shoot in" the weapon and test all of its parts, part of the process of acceptance for delivery. Could be as low as 0 rounds, could be as high as 20 or 30 or more on lighter weapons (37mm for example). The reason was to vibrate the parts as see if everything was functional and still working. For the same reason, the a tank was driven about and the engine retuned, and a rifle was shot and then taken apart etc.

I don't know anything about quandrant sight testing for acceptance and ammunition expenditure in the testing process, but am interesting the the acceptance process if you are willing to tell more. How are quandrant sights tested with the weapon when the weapon is accepted into military use in modern practice? When you were processing weapons into the military, were quadrant sights tested for acceptance apart or with the weapon system? Do replacement quadrant sights accepted by the logistic chain from manufacturer get tested away from the weapon or on it? Is there some ranking system for the tests or is it pass / fail. In modern use, if weapons are tested for acceptance with live ammunition, does anyone place the quadrant sights onto the cannon?

I have some other things that I am confused about. Since my sources say that the Tiger needed to have its sights brought into alignment with the tube on a regular basis, and artillery does not, would these not be different issues all together, or do you quadrants work both as direct fire or indirect fire elements? How similar is a Tiger tank to the artillery piece of your experience in maintinence and operation? If they are different, how are they different?

[ 10-04-2001: Message edited by: Slapdragon ]

[Wreck]

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

Surely it makes no tactical difference to him whether the Tac AI fires 35 rounds at empty space and hits one tank with the next 5, or fires 40 to hit one tank.

If those two cases don't make any difference to you, then I have to conclude we have to agree to disagree here as continuing this discussion would be pointless to the extreme.

<HR></BLOCKQUOTE>

Tactical difference. I see no tactical difference, do you? 40 shots fired, one kill.

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

Note that I'm not demanding that CM tanks should fire at non-existing targets. While being arguably more realistic, it would be a mistake from playability viewpoint.

<HR></BLOCKQUOTE>

If the effect is really as great as some here seem to suggest, then leaving it out of CM is a mistake. If most of what tanks historically did shoot at was bushes, then they should do so in CM.

This is not that hard to do. Just make the tac AI target "unidentified vehicle" markers, and then change the engine to generate lots of fake ones. To get a 19/20 ratio of shots fired at empty space, I would suggest about 10 fake targets per real enemy tank would be about right, assuming it takes two shots per fake to eliminate. Fake targets would come into existence in scattered trees, brush, and near buildings, and would only be dispersable by fire or better observation.

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

However, arguing that those rounds that would have been wasted in a real combat should be factored in by lowering general accuracy, would be in my opinion quite strange.

<HR></BLOCKQUOTE>

Strange, yes. The effect is not exactly like what I propose above. However, the effect does fairly closely model the assumed reality, namely, that tanks need to carry a lot of "extra" ammo to fire at bushes, in order to have a sufficient amount for their one or two shot kills on the real vehicles.

If you fail to model the fact that 95% of rounds are being shot into the ether, then you make tanks into something they were not. You make them able to kill 40 times their number, when in reality they would only be likely to be able to kill 2. Multiplying tank lethality by 20 is an effect that is bound to strongly degrade the accuracy of the simulation.

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

Where lo, they will perform the exact same tactical role as misses, because misses by any other name is what they are.

<HR></BLOCKQUOTE>

Actually the effect is not the same; in order to shoot bushes the tank needs to point its gun at them. This may well lead to flank shots from the real enemy targets that would not otherwise happen.

Be that as it is, though, the effect is pretty similar.

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

Do you think that CM should artificially lower the accuracy of targeted shots so that the average ratio of shots/kill is close to whole-war average?

<HR></BLOCKQUOTE>

No of course not, but that is because CM does not need or try to model accurately all of the combat that happened. CM models reasonably even fights, at fairly close range, with fairly good visibility. It does not need to model accurately things like a platoon of Tigers on a hill that happen to see a platoon of Shermans on an open road below at 2000m, and open fire killing them all. It also does not model night combats well, the sort where 95% shooting at bushes might be believable.

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

My viewpoint is that in tactical combat it is more important to get the accuracy of correctly targeted shots correct. Your mileage may vary.

<HR></BLOCKQUOTE>

All this tells me is that in your heart, you don't think that the amount of bush-shooting that went on in tactical combat was all that large. Neither do I.

[tss]

Wreck wrote:

Tactical difference. I see no tactical difference, do you? 40 shots fired, one kill.

Actually, I do. Suppose that an enemy tank pops out to view and you start shooting at it. It makes a significant difference whether you hit it with your third round or sixth, since in the time those three extra shots take the target may get back to cover again.

So, in this example, if wasted shots are not factored in accuracy, there's a dead enemy tank. If not, there's a live enemy tank.

I agree that strategically there's no big difference, but tactically there is.

If the effect is really as great as some here seem to suggest, then leaving it out of CM is a mistake.

I don't know how large effect it was. I don't have enough data to determine that.

If most of what tanks historically did shoot at was bushes, then they should do so in CM.

Well, it could also be said that since infantry often (with a suitable value of "often") stopped for 15-30 minutes while the platoon leader tried to get some picture of what was happening around, they should do so in CM.

This is not that hard to do. Just make the tac AI target "unidentified vehicle" markers, and then change the engine to generate lots of fake ones.

That would be actually a neat thing to have, also with infantry units. It would certainly add a factor of uncertainty to attacking.

You make them able to kill 40 times their number, when in reality they would only be likely to be able to kill 2.

On the other hand, the number of possible kills in a scenario is bounded from above by the number of enemy tanks present in the battle so the situation is not quite that severe.

One thing could be that tanks started the battle without a full ammo load. Though, I'm not certain whether that would really be a good idea.

All this tells me is that in your heart, you don't think that the amount of bush-shooting that went on in tactical combat was all that large. Neither do I.

I would like if I could give a reasonable estimate on the size of the effect, but I can't. As far as I know, it might have been that 90% as in my above-mentioned example. Or it might have been as low as 5%. Or something between them. Or something outside the range. My gut feeling is that the effect was significant but it wouldn't be the first time that my intuition gives an incorrect answer.

- Tommi