There is hope for the Sherman

[Rocky Balboa]

Tis a sad day indeed when accounts of those who actually fought in a conflict are dismissed in favour of theoretical combat models, which are themselves idealised capabilities. Not every shell travelled at a standard velocity as propellant technology was not a perfect science, especially quality control, not every piece of steel plate had standard properties etc. British WWII research stated that there were so many factors influencing penetration of armour that any performance figures for guns was an approximation at best.

DT have Tank Men but need to get time to read it, any good, most people seem to think it is ok but could have been better, the proverbial curates egg (modern usage not the original from Punch)

No simulation is perfect just as no model based soley on historical account will be perfect. The point here is which will be less perfect? It's a very iffy thing to model based on historical account because you have to code using all kinds of abstractions so that it feels the way someone thinks it should feel. BFC decided to take the route of letting the numbers tell the story and I find it refreshing.

Could or did the numbers vary? Of course I'm sure they did, but how much? As BFC moves forward, if it can be proven that armor quality or shell propellant varied greatly from the same tank model of the same design then I'm sure they can build in some variability to their base numbers.

Of course when you do this then you are engineering more chaos into your simulation which is exactly the thing you are arguing against. The nasty thing about chaos is that you must be willing to accept the good results with the bad.

[gunnergoz]

I suspect that BFC gives more credit to veteran accounts than many might guess, but they are also very scientifically minded and data-driven which makes them temper their results based upon the latter, slightly adjusted to account for the former when possible.

[Dietrich]

The other day a Sherman of mine tht was hull-down behind a ridge took two hits from a PaK 40 to its right front turret which ricocheted. The Sherman took only slight damage, but the shell fragments took out half of the men in a squad of mine kneeling behind the bocage 20 feet away. (I've even seen AP shells richocet intact and arc away to hit friendly infantry more than 50 meters away. Talk about detailed simulation!)

[womble]

The other day a Sherman of mine tht was hull-down behind a ridge took two hits from a PaK 40 to its right front turret which ricocheted. The Sherman took only slight damage, but the shell fragments took out half of the men in a squad of mine kneeling behind the bocage 20 feet away. (I've even seen AP shells richocet intact and arc away to hit friendly infantry more than 50 meters away. Talk about detailed simulation!)

Heh, I've not seen any 'collateral damage' of that kind from ATG hits yet (though a 'saved by the woodwork' moment did kill two dogfaces 20m and a wall from a tank; they weren't even under the tree's canopy... ) but I'll certainly concur that Shermans seem pretty resilient against the ol' Pak 40. So far, I think it's taken about 10-12 hits on my Shermans, across the games I've played, for a Pak40 to claim one tank kill. Some yellow crew scratches, but just the one forced to abandon. And the 50mm haven't been able to keep it in their shorts long enough to get flank shots...

[Battlefront.com]

We have to remember that the Pak40 was introduced in large numbers in 1942. By 1944 it was no longer the primary armament for the Germans. There's a reason for this, just as there is a reason it replaced the Pak38 after barely 2 years in service. WW2 was so fast moving from a technology standpoint. Pretty much every 6 months something came out to trump whatever was average in the previous 6 months.

Steve

[gunnergoz]

Yes, what makes it even more interesting is that, for much of the war, while the Germans and UK were playing high-stakes Blackjack, the US was playing Bridge...

[Bigduke6]

Unfortunately, Human recollection can be really bad. At least when it comes to details. It's why eye witness testimony is seen as being unreliable. Even when corroborated by other eye witnesses. Do some net searches on this and you'll find some really stunning studies done on the topic. Especially for/by law enforcement.

Should we dismiss anecdotal and personal combat results completely? No, absolutely not. They are often extremely important. But they are also extremely unreliable on the whole without other evidence to back them up. They are also useless from a statistical standpoint.

Incorrect. Ballistic performance is pretty close to an exact science. There can be a HUGE range of results, of course, but they can be mathematically explained. On the other hand, the "soft factors" (Human factors) that go into a particular shot and hit are wildly varied and with a healthy dose of randomness. This is something that is semi-scientific, but ultimately judged by outcomes. CM excels at this sort of modeling.

One of the very, very frustrating things about official reports (the Brits did a great one right after Normandy) is that they tend to treat all Shermans as if they were identical. As the game shows, there are HUGE differences between the models. The Brits, for example, used a lot of older models that the Americans did not use. And those were quite a bit more vulnerable to 75mm fire by comparison. So those stories of 1 hit kills of Shermans could definitely be accurate in the right context (i.e. North Africa, Italy, and British equipped forces in France). The Brits also didn't start out in France with many Wet Stowage tanks, so they had more problems with burnouts than did the Americans. etc. etc.

Steve

Steve,

I've bolded the part of your comments I disagree with. I think that anecdotal evidence can be useful in statistical analysis. Primarily, as a reality check if you build a statistical model of a historical event, like for instance WW2 armored combat.

Certainly anecdotal evidence can't be used as raw data on which to build a statistical model, in the first place because humans under stress perceive things in really variable ways, and in the second place because the anecdotal data available can be pretty far from a fair sampling of the actual reality. For instance, anecdotal information is if nothing else skewed towards the experience of survivors.

However, and not every one agrees with this, in my opinion there is such a thing as the weight of anecdotal evidence. If there are enough anecdotes saying the same thing, and if a responsible check of the evidence seems to bear out that the anecdotes are consistent among themselves and with other historical evidence, then that is the yardstick by which one determines historical "fact".

This becomes more important when one considers that although modeling AP to armor engagements is a science that can be reduced down to mathematical models, modeling them accurately for a computer is not as easy. This is because, in the real deal, the engagements were not AT cannon crews firing at defenders holding up plates of armor in front of themselves inside a test lab.

Weather, tube motion, shell-to-shell quality, barrel wear, maintenance of both gun and shell, etc. etc. all could influence an individual shell's performance. Maintenance, age, junk hung on the outside, relative proximity to a welding seam, batch-to-batch quality, previous AP shell strikes, movement, etc. etc. could influence the armor's performance. These variables are, quality aside and as far as I know, not modeled in the game. This does not make the simulation inaccurate. Far from it. But it does make reasonable checking the simulation's modeling of AP shell to armor engagements against the historical record.

Indeed, the historical record provides useful information, and sometimes it allows us to jettison anecdotes. I is easy to reject historical anecdotal evidence that most panzers encountered by Allied troops were Tigers, and most AT weapons fired at them were 88s. Why? Because we have more solid data that clearly contradicts that anecdotal evidence, to wit the German military historical record; we know how many Tigers were produced and where they were fielded and in what numbers. Given that, dismissing the Allied "truism" that every panzer was a Tiger and every cannon an 88, in Normandy, is easy. The game accurately reflects that with its - in my opinion - brilliant rarity point system. Whoever thought of that is one smart guy.

That said, it seems there is some evidence that CMBN models the Allied 75mm AP shell as unable to penetrate the side or rear of a Tiger I from extremely close range. I'm personally not 100 per cent positive this is the case, but for the moment let's assume it is.

Let's further assume - and I think this is an excellent assumption - that CMBN models the 75mm AP shell engaging the Tiger I's 82mm armor extremely well, i.e., probably better than any other computer simulation anywhere, by anybody. Does this mean we can safely ignore strong anecdotal evidence, that the only way Sherman crews could deal with a Tiger, is by getting close and hitting it from the flank and rear?

I would say "no". This set of tactics was after all a tactical doctrine across the entire Allied force, and it was based on a historical understanding of what the Sherman's 75mm could and could not do against a Tiger. The Allied force, after all, had an advantage over CMBN, they had real captured Tigers and real 75mm AP they could test against them. Their conclusion seems to have been "Get close, hit the Tiger in the flank or rear."

Are we to dismiss that in favor of CMBN modeling of that same engagement, which - again, assuming 75mm AP performance does not overcome Tiger I's 82mm armor - produces a very different doctrine: An Allied player should never engage a Tiger with Shermans, period, at any range?

I think I have an idea of the Pandora's Box that gets opened, if one starts going into the game engine and mucking around with its AP shell-to-armor mechanics, for the sake of producing a tactical outcome similar to the predicted result in the real deal. Maybe it's just too hard to do, I just don't know.

But one thing seems clear to me: If you have enough anecdotes, and they are borne out by other evidence, that can indeed be a useful check on a statistical model. Although if the check shows a problem - and to repeat myself, I'm aware I may well be going on here about a straw man - it's probably not going to be so easy to do something about the problem.

I'm sure that if Tiger sides and rears provide to be tougher in the game than in the real deal, it will become clear enough quickly enough.