Differing views of Overlord

[Carl Schwamberger]

...

Can you elaborate on that one? There was a large amount of concrete poured in the latter 1943, then winter & spring of 1944, not just along the possible landing sites , but around the ports as well. What specifically are you looking at that cancels the difference is the quantity of concrete protecting the defenders? Beyond that this seems to strongly connect to the difference in defense strategy & expected operational & tactical methods between 1943 & 1944 and the changes in personalities at several levels of command.

It's an interesting sidelight to NEPTUNE, the German analysis of the effectiveness of the building program. The most interesting finding was that adding the overhead cover to the coast artillery batteries may have been counterproductive. The problem was the "protected" emplacements were more easily disabled than the older open emplacements because the embrasure tended to "funnel" naval counterbattery rounds into the gun house where they exploded with devastating effect. Marcouf was an example, but apparently others suffered similarly. Since the gun crews tended to remain in the gun house, they were killed or injured, and the concentration of the blast and fragments did more damage to the guns and accessories, but in the open emplacements the crew shelters were separate, protecting the crews and only a direct hit on the gun itself tended to damage it.

There would not be the flanking beach casemates, but I suspect that field fortifications in enfilade would have been as effective.

This is a very narrow bit of ‘data’.

Funneling naval cannon shots into embrasures, or of any cannon, is a long and tedious thing. Unless you are at very close range & have precise identification of the target that sort of accuracy is problematic. Very close range would be under 1200 meters for lighter cannon like a tank gun, a bit more for a 5” naval gun, or maybe between 1500 & 2000 for larger naval cannon. Thats great when your tank, or destroyer playing chicken with the surf is supporting a assault team, but for general fire support the odds of hitting a embrasure is too small to consider. It happens, but never count on it.

In this larger context the concrete has the advantage over the sandbag and log structure in that the steel reinforced concrete is far more resistant to the hit on the other 90%of the structure other than the embrasure. I can spend several days digging out the data of shot vs assorted protective cover, but I suspect you have some of that near your comfy chair, so I dont need t reproduce your data. Effects on target can be displayed as a spectrum across scale of protection. The harder the material the more ammunition it takes to gain effect. Or for a practical quantity of ammunition it moves the lesser effect along the scale. For the duration of the sort of preparatory fires there were on the Normandy beaches concrete moved the effects to the far end of the scale to Suppresive. Against lght entrenchemnts the effct vs infantry and cannon defense would have been near the Destructive end of the scale. Log or steel & sandbag structures turn identical fires to the Nuetralizing & Suppresive portion of the scale. Concrete & steel shifts the effect firmly into the Suppresive zone.

Beyond that theres the many other details differing between the two years, the quantity of mines, wire, AT ditches, communications trenches, MG, AT guns, field artillery, men, buried telephone wire, ect... ect... Through the advent of Rommel the German strategy was to not defend the beaches. It was to defend the ports, denying those to the enemy & defeat the enemy force by concentrating substantial reserves for a counter attack. Most folks have at least a cursory understanding of this. What they miss is the relatively thin line of outposts that covered the beaches & lack of local reserves. Before Rommel think of Nettuno Italy covered by a company deployed in a observation role, and a couple battalions within six to twelve hours march.

I could go on at length but need to get to work. I think that any more than a cursory glance at the German ground force shows just how weak it was man for man vs what Rundstead had in May 1944.

[Richard Anderson]

True, up to a point.

Yep, up to the point that it is judged proprietary.

However to outsiders it is a 'black box' as the exact results cannot be replicated by third parties. I am not aware that the values attributed to weapon systems (or the modifiers for combat conditions) are publicly available. Therefore even detailed research on individual engagements to identify participants, weapon systems and results will not necessarily come up with a similar answer, comparing "expected" and actual results.

Really? They cannot? Anyone that runs an engagement with the same inputs gets the same output.

If you want to argue about the research and choices in the engagements, then that is a different subject and has nothing to do with the model.

In theory the engagements used shouldn't greatly impact the results if all factors have been correctly included and weighted in the model. Again without transparency, this is difficult to prove.

Of course the choice of engagements doesn't effect the results. The choice of inputs does.

The model has also evolved over time as shown by the multiple (rev) in the list of engagements, so I suspect some past criticisms might have been correct.

Sigh...those are engagement datasets, not model runs. The revisions (rev) are all from a review of engagements data we did about twenty years ago.

I'd love to get my hands on the model to do some what-ifs, both serious and less so eg

Between NPW, Attrition, and Understanding Defeat you can create a paper and pencil version, which many people have done. It is essentially the same as the QJM, which computerized the calculations. The TNDM reprogrammed the model and used differential equations for casualty calculations created by Dr. Jim Taylor of the NPGS.

How does combat power in units fall as casualties are incurred - I suspect initially slowly, then a rapid drop-off at some point?

Loss of weapons systems is the largest factor in combat power decreasing and is linked to personnel casualties. The other factor affecting combat power is time in combat.

What impact does phase of the moon have?

It substantially reduces the effectiveness of SS Werewolves.

Unfortunately I'm not prepared to pay for it, so I remain sceptical rather than critical.

Seriously, you can experiment with the modeling very easily with paper and pencil...it is just a long mathematical process.

So what are the real problems with the model rather than the imaginary fictitious ones?

1. It was only ever really extensively tested against "divisional-level" engagements during the period 1939-1945 and 1967-1973. So "classic" mass-army, mechanized warfare of the 20th century. It may lose fidelity for much smaller and much larger engagements.

2. It does not well replicate certain types of engagements. Don't try to replicate D-Day on OMAHA for example...the Americans will steamroller over the German defense since it cannot replicate the limitations of the NGFS and the phased arrival of the various landing waves.

3. It is - like all models - highly dependent on the analyst in setting up the engagement parameters.

4. The model interface itself is 20 years old and difficult to use properly.

[Richard Anderson]

...

Not out there enough. Better to discuss a all British operation executed a few days after Op BARBAROSA starts. Whats a good high tide date between 22 June & 6 July 1941.

Seriously? Read Alanbroke's diary - the unexpurgated one - never gonna happen.

Absolutely not seriously. That was a joke son. Lighten up

Thank God, I was seriously worrying about your mental state...or possibly the degree of your inebriation.

[Richard Anderson]

... The other project I wish I could have completed - it was never really funded - was to finish the dekade campaign data base for North Africa and Italy, since we have almost 100% complete German casualty data (personnel and major items of equipment) from February 1941-May 1943 and September 1943-June 1944 and could match it to allied data...

Out of curiosity, what does such a project cost? I do watch for lottery tickets on the sidewalk & might get lucky.

IIRC, the three years of work on the Ardennes data base ended up costing around $.5 million? That was 30 years ago, so likely triple that. I don't recall what the Kursk data base, also about three years effort, ended up? Probably about $1 million? Small change compared to the budgets of outfits like RAND. The real problem is there is virtually no interest in defense circles for such work anymore because of the perception that conventional warfare is "legacy" and/or that "modern" warfare if it is "conventional" will be "different" because of computers.

[Tom from Cornwall]

2. It does not well replicate certain types of engagements. Don't try to replicate D-Day on OMAHA for example...the Americans will steamroller over the German defense since it cannot replicate the limitations of the NGFS and the phased arrival of the various landing waves.

Which is one of my points about the Salerno beachhead engagement scenarios.

Paestum Beachead
Amphitheater Beachead
Port of Salerno Beachhead
Port of Salerno (rev)

Infantry units, artillery, armour and engineers all tumbling ashore in the right or wrong order and place and not necessarily actually making it ashore at all as coherent formations and, in some cases I expect, not really operational until D+1.

Relevant to the thread title is a look at footage of the British beaches of Salerno on D-day (9 Sep 43) and comparing that with the footage of the beaches of Normandy on that D-day. The extraordinary increase in armoured support, SP guns, SP AT guns, armoured engineer vehicles, carriers, armoured cars, SP LAA guns, etc shows what was learned from landing insufficiently supported infantry formationS against the threat from early German armour counter-attacks.

Regards

Tom

[Richard Anderson]

This is a very narrow bit of ‘data’.

Indeed. The real question is whether or not it was simply German perception of a problem with the design of their emplacements or whether it was an actual fault in the design. The best evidence that I have seen is that it was a fault in the design.

Funneling naval cannon shots into embrasures, or of any cannon, is a long and tedious thing. Unless you are at very close range & have precise identification of the target that sort of accuracy is problematic. Very close range would be under 1200 meters for lighter cannon like a tank gun, a bit more for a 5” naval gun, or maybe between 1500 & 2000 for larger naval cannon. Thats great when your tank, or destroyer playing chicken with the surf is supporting a assault team, but for general fire support the odds of hitting a embrasure is too small to consider. It happens, but never count on it.

Except it was not something attempted by the NFGS. With few exceptions they fired at designated predetermined target areas rather than at point targets - the destroyer actions at OMAHA were the exception rather than the rule. And yet the result was the two major fully protected "modernized" batteries - Marcouf and Longues-sur-Mer - were easily knocked out by relatively few rounds, while the other "older" open emplacements survived bombardment after bombardment.

In this larger context the concrete has the advantage over the sandbag and log structure in that the steel reinforced concrete is far more resistant to the hit on the other 90%of the structure other than the embrasure.

Indeed...the problem is I'm not talking about "sandbag and log" structures. The "older" emplacements were open concrete structures, conventional coast artillery emplacement designs similar to those used by the U.S. and Britain in pre-war emplacements. Worse, the finding was that even if 90% of the round struck the structure, the tenth that hit the embrasure resulted in a hard kill of the gun and crew.

I can spend several days digging out the data of shot vs assorted protective cover, but I suspect you have some of that near your comfy chair, so I dont need t reproduce your data. Effects on target can be displayed as a spectrum across scale of protection. The harder the material the more ammunition it takes to gain effect. Or for a practical quantity of ammunition it moves the lesser effect along the scale. For the duration of the sort of preparatory fires there were on the Normandy beaches concrete moved the effects to the far end of the scale to Suppresive. Against lght entrenchemnts the effct vs infantry and cannon defense would have been near the Destructive end of the scale. Log or steel & sandbag structures turn identical fires to the Nuetralizing & Suppresive portion of the scale. Concrete & steel shifts the effect firmly into the Suppresive zone.

That is true for the emplacements in defilade. It was not true of the emplacements that could not be placed in defilade, such as the coast artillery.

Beyond that theres the many other details differing between the two years, the quantity of mines, wire, AT ditches, communications trenches, MG, AT guns, field artillery, men, buried telephone wire, ect... ect... Through the advent of Rommel the German strategy was to not defend the beaches. It was to defend the ports, denying those to the enemy & defeat the enemy force by concentrating substantial reserves for a counter attack. Most folks have at least a cursory understanding of this. What they miss is the relatively thin line of outposts that covered the beaches & lack of local reserves. Before Rommel think of Nettuno Italy covered by a company deployed in a observation role, and a couple battalions within six to twelve hours march.

The major factor actually missing from that equation in 1943 versus 1944 is the seaward obstacle belt, the minefields, and many of the concrete emplacements - Rommel's "devil's garden". The idea that the "beaches were not defended" until Rommel arrived and "changed" the German strategy falls down on examination. The divisions were there and defending the beaches - the coastal crust always existed, although it was improved.

The question then becomes what happens when a less sophisticated and powerful Allied invasion force attempts to land on those beaches? The real problem I suspect would be the Allied inability to rapidly reinforce the initial mostly infantry force and the inability of the Allied air forces to interdict German reinforcements to the battle area.

I could go on at length but need to get to work. I think that any more than a cursory glance at the German ground force shows just how weak it was man for man vs what Rundstead had in May 1944.

Okay, more repetition from 14 years ago...

Number (incl. Schnell) June 1943/Number (incl. Schnell) June 1944
Total Western Front 60 (12)/91 (20)

In terms of degrees of readiness, 6 of the 8 Schnell divisions in France in 1943 were either ready or nearly ready by 1 June 1943 (75%), although a lot of their assigned tanks were still in transit. In contrast in 1944 only 5 of the 10 Schnell divisions in France were either ready or nearly ready by June 1943 (50%).

Of the other 39 divisions in France in 1943, 23 were infantry (including, 5 newly formed or reformed and 5 static), 1 was SS Grenadier, 10 were Reserve-Infantry, 2 were Fallschirm, and 3 were LW-Feld, so about 18 of the 39 (46%) were 'effective'. In 1944 the other 43 divisions included 31 Infantry (including 2 “reflagged” Reserve-Infantry, 14 newly formed, and 4 static), 7 Reserve-Infantry, 3 Fallschirm (all newly formed), and 2 LW-Feld, so about 13 of the 43 (30%).

So in 1943, a force about one-third smaller, but slightly more effective overall.

As of June 43 available for European operations (ETO and MTO), US forces would have had the following:

2 armor divisions
6 infantry divisions
1 airborne division

Compared to June 44:

8 armor divisions
21 infantry divisions
2 (reinforced) airborne divisions

British forces:

June 43:

8 armor divisions
18 infantry divisions
1 airborne division

June 44:

Effectively unchanged.

Note however, that a proportion of those forces would have to remain in North Africa under any realistic scenario. Assume no TORCH, then assume Rommel recovers in Tripoli and thus there is a new stalemate, taking c. 2 armour and 5 divisions out of the possible British OB for 1943. So call it 29 Allied divisions in 1943, compared to 58.

In essence, 66% of the German force in 1943 would be facing half the Allied force.

However, what is more important is the rate of possible build-up. The build-up of follow-on divisions June to September 43 in the ETO would average one per month, it is unlikely that the Germans would have been unable to match or exceed that rate of buildup. The rate would increase to two divisions per month for the US from September to December 43, a rate that the Germans should have had no problem matching or exceeding.

[Mori]

So what are the real problems with the model rather than the imaginary fictitious ones?

1. It was only ever really extensively tested against "divisional-level" engagements during the period 1939-1945 and 1967-1973. So "classic" mass-army, mechanized warfare of the 20th century. It may lose fidelity for much smaller and much larger engagements.

2. It does not well replicate certain types of engagements. Don't try to replicate D-Day on OMAHA for example...the Americans will steamroller over the German defense since it cannot replicate the limitations of the NGFS and the phased arrival of the various landing waves.

3. It is - like all models - highly dependent on the analyst in setting up the engagement parameters.

4. The model interface itself is 20 years old and difficult to use properly.

And a 5th item:

It is a pretty comprehensive sample of available data, although by its very nature it could not be randomly collected, since the people paying for it specified the collection parameters. Versteht?

Since it's far from a random sample of engagements, no overall conclusion on combat effectiveness of Germans vs. US or Germans vs. British can be derived from it.

This is what Dupuy once wrote, and it would only stand had he picked a random sample, which wasn't the case:

On a man for man basis, German ground soldiers consistently inflicted casualties at about a 50 percent higher rate than they incurred from the opposing British and American troops under all circumstances. This was true when they were attacking and when they were defending, when they had a local numerical superiority and when, as was usually the case, they were outnumbered, when they had air superiority and when they did not, when they won and when they lost.

[Richard Anderson]

Which is one of my points about the Salerno beachhead engagement scenarios.

Infantry units, artillery, armour and engineers all tumbling ashore in the right or wrong order and place and not necessarily actually making it ashore at all as coherent formations and, in some cases I expect, not really operational until D+1.

Which is why those were not actually run as beachhead assaults, although I attempted to do so at one point with mixed success. The problem with the QJM/TNDM I forgot to add was that it is effectively normalized as engagement "days" being 24 hours. When you try to parse a day into say the three to four hours of the actual beach assault the results get wonky. So those engagements are not just the assault phase, but the result of the entire day.

The confusion remains though. Those are not records of engagement runs, they are records of the historical engagements.

Relevant to the thread title is a look at footage of the British beaches of Salerno on D-day (9 Sep 43) and comparing that with the footage of the beaches of Normandy on that D-day. The extraordinary increase in armoured support, SP guns, SP AT guns, armoured engineer vehicles, carriers, armoured cars, SP LAA guns, etc shows what was learned from landing insufficiently supported infantry formationS against the threat from early German armour counter-attacks.

Oh, but of course they wouldn't need that in June 1943 because the Germans didn't have an armor to counter-attack with...or something like that.

In fact, German armor strength as of 31 May 1943:

Total for OB-West – Panzer 140/31/287, StuG 53/0/50, Beute-Panzer 317/83/0
Total for OB-Sued – Panzer 279/12/70, StuG/StuH 169/2/51
Total in the west – Panzer 419/43/357, StuG/StuH 222/2/101,
Beute-Panzer 317/83/0 (1st Line = 641/45/458, 2nd Line = 317/83/0)

As of 31 June 1944:

Total in OB-West – PzII 16/4/0, PzIII 71/20/9, PzIV 667/92/124, PzV 444/99/278, PzVI 46/7/51, Total Panzer 1244/222/462, StuG/StuH
318/27/10
Total in OB-Suedwest (formerly Sued) – PzII 2/2/0, PzIII 85/27/29,
PzIV 210/62/102, PzV 54/16/38, PzVI 81/9/33,
Total Panzer 432/116/202, StuG/StuH 341/111/62
Total in the west – Panzer 1676/338/668, StuG/StuH 659/138/72

So notably weaker in 1943, but then the ability of the Allies to land armored units is also much less.

For American forces the armor available was roughly:

1943 - Total = 788 medium tanks + 197 reserve (est.), 532 light tanks + 133 reserve (est.), 96 M3/M8 assault guns + 24 reserve (est.)
1944 - Total = 2,964 medium + 851 in reserve (est.), 1,507 light + 346 in reserve (est.), 286 M8 assault guns (TE) + 70 reserve (est.)

The real problem though was the lack of sufficient LCT and LST to get them ashore in a timely manner.

[Richard Anderson]

And a 5th item:

It is a pretty comprehensive sample of available data, although by its very nature it could not be randomly collected, since the people paying for it specified the collection parameters. Versteht?

Since it's far from a random sample of engagements, no overall conclusion on combat effectiveness of Germans vs. US or Germans vs. British can be derived from it.

Sure. Did I say any different? Tentative overall inferences may be made, but it is far from conclusive, which is why Trevor - and we after his death - always said there needed to be more research and testing. The result? Crickets. No one in the defense or defence establishment gives a damn.

This is what Dupuy once wrote, and it would only stand had he picked a random sample, which wasn't the case:

On a man for man basis, German ground soldiers consistently inflicted casualties at about a 50 percent higher rate than they incurred from the opposing British and American troops under all circumstances. This was true when they were attacking and when they were defending, when they had a local numerical superiority and when, as was usually the case, they were outnumbered, when they had air superiority and when they did not, when they won and when they lost.

Why yes, it was consistently true among those engagements examined and consistently true in the engagements examined after he wrote that. Thus, it stands for that non-random sample. Now find those random examples were it doesn't stand.

[Mori]

Why yes, it was consistently true among those engagements examined and consistently true in the engagements examined after he wrote that. Thus, it stands for that non-random sample. Now find those random examples were it doesn't stand.

Well, Dupuy should have moderated his conclusion by reminding readers he had left aside situations similar to the ones pointed above (7th US Army, Veritable-Grenade and everything from the 1945 fight), situations which were sizable in terms of division-days and which happened to mostly be US / Allied domination over their German opponents,

That's my only point.

And, if we go back to what started the discussion, I believe this is why Dupuy and al. have been gradually ignored by authors of military history.

As a side remark, I have noticed that undergrad university courses now include solid classes on statistics for most humanities (sociology, history, ethnography, linguistics, but also business schools, medical studies etc.). 20-30 yrs ago, stats were something touched upon in mathematics/physics curriculum, but it was exceptional to see a stats course elsewhere. I would put as an hypothesis that the limits of a 'non-random sample' or of a statistical study without hypothesis testing have become common knowledge among historians younger than 40. Historians are now trained to see these problems.

[Aber]

Seriously, you can experiment with the modeling very easily with paper and pencil...it is just a long mathematical process.

I have NPW and after looking at a sample engagement realised that without the weightings of weapons, and the modifying factors for conditions, I could not replicate the calculations. I could of course create my own model from scratch if someone paid me.

[Richard Anderson]

Well, Dupuy should have moderated his conclusion by reminding readers he had left aside situations similar to the ones pointed above (7th US Army, Veritable-Grenade and everything from the 1945 fight), situations which were sizable in terms of division-days and which happened to mostly be US / Allied domination over their German opponents,

Um, why? He was quite clear throughout that the results were for that specific data set. The inference that it applied in general is obvious, but then it is an inference that is supported by the data.

You are now making an unsupported inference regarding "7th US Army, Veritable-Grenade and everything from the 1945 fight". You have no idea what the force ratios or casualty ratios were, so do not know if the Allies "dominated" or simply overwhelmed their German opponents. Given that I suspect from the force ratios I could suss out that the Germans were overwhelmingly outnumbered and outgunned in those cases, it is hardly unsurprising they were defeated, but that gives us zero evidence that Trevor's inference drawn from the existing data set did not hold true.

And please be clear. I am not saying that the "50%" factor remained constant in those cases. I'm saying WE DON'T KNOW and WE CAN'T KNOW either way. Their is insufficient data to test and draw ANY "consistent" conclusions. It is not a problem of "non random data" it is a problem of nonexistent data.

Furthermore, the best evidence that can be drawn from the limited data we do have is that - unsurprisingly - the German advantage was waning over time, albeit slowly as they scraped the bottom of the manpower barrel, as their ability to fire and maneuver decreased along with their supplies of fuel and munitions, and as Allied capabilities and experience increased.

What a surprise.

That's my only point.

Good to know.

And, if we go back to what started the discussion, I believe this is why Dupuy and al. have been gradually ignored by authors of military history.

Not ignored, more like unknown, and uninterested in such analysis. Your average "historian" has little knowledge of or interest in OR analysis, its historiography, or its significance. Try to find references to Millet and Murray, AORG, RAC, or anything similar in ANY current military history. Or, on the other side of the coin, look at the occasional ignorance of history on the part of the OR types - Millet and Murray in A War to be Won buying wholesale into the ahistorical narrative of Belton Cooper's Death Traps...or Trevor's famous (to those aware of it) battle that never was.

As a side remark, I have noticed that undergrad university courses now include solid classes on statistics for most humanities (sociology, history, ethnography, linguistics, but also business schools, medical studies etc.). 20-30 yrs ago, stats were something touched upon in mathematics/physics curriculum, but it was exceptional to see a stats course elsewhere. I would put as an hypothesis that the limits of a 'non-random sample' or of a statistical study without hypothesis testing have become common knowledge among historians younger than 40. Historians are now trained to see these problems.

Sure, whatever you want to believe... Seriously, adding or subtracting classes on various things just adds to the classes that need to be checked off to obtain a degree, it does not increase knowledge, it increases education. Not exactly the same thing. It is all too obvious - to me at least - that adding a stats course will do nothing to help historians to find where problems in their assumptions are.

[Richard Anderson]

Seriously, you can experiment with the modeling very easily with paper and pencil...it is just a long mathematical process.

I have NPW and after looking at a sample engagement realised that without the weightings of weapons, and the modifying factors for conditions, I could not replicate the calculations. I could of course create my own model from scratch if someone paid me.

That's all that you need for basic QJM calculations. Weapons scores in the QJM/TNDM are not weighted, unless I'm missing what you mean. An OLI is an OLI whether it is infantry, armor, or artillery. The modifying conditions are terrain, weather, etc...and are all there. They were later refined and tweaked a bit, but the basic calculations remain the same.

[Tom from Cornwall]

The real problem though was the lack of sufficient LCT and LST to get them ashore in a timely manner.

And of course the lack of air superiority to ensure that whatever LCTs and LSTs etc there were got to the 'far shore' without the Germans knowing 24 - 48 hours in advance that an invasion was on its way and of attacking the invasion force at sea.

Which is why those were not actually run as beachhead assaults, although I attempted to do so at one point with mixed success. The problem with the QJM/TNDM I forgot to add was that it is effectively normalized as engagement "days" being 24 hours. When you try to parse a day into say the three to four hours of the actual beach assault the results get wonky. So those engagements are not just the assault phase, but the result of the entire day.

I'm still not sure, from what you say, that enough allowance is made for the disorganisation inherently inflicted on a formation by the chaos and confusion of an amphibious landing - I think it probably took several days for a division launched ashore to become equal once more to the sum of its parts.

I'm happy to agree to disagree though...

BTW there is an argument (but I wouldn't dare make it!) that British formations performed better under 8th Army rather than under 5th Army in Italy. All to do with the supply of tea and rum I expect. And of course the language barrier...

Regards

Tom

[Aber]

An OLI is an OLI whether it is infantry, armor, or artillery.

And IIRC NPW gives OLI for only a limited set of weapons - 60mm mortar, 105mm howitzer, Sherman tank, and P-47 for WW2.