Daily advance rate comparisons

[Qvist]

The distance between the starting point of LVI. AK and Daugavpils is not 310 km as the crow flies, it is about 310km via road, today. They did not start from Tauroggen, but from east of Tilsit, and then went east immediately via Jubarkas. Your result of 478 is a bug in Viamichelin, or maybe related to current road closures (check the map, you'll see what I mean). Considering the very close match between the road distance and the distance used by Z/F, I have a hard time believing that what they used is aerial distance in all cases. The very small differential between the Brest-Slonim-Bobruisk figures seems to confirm my impression.

Well, you are wrong.

LVI:
Yes and no. For some reason the software makes a detour down to Vilnius and then up again, clearly a bug. But if we use Tilsit-Jubarkas, and Jubarkas-Kaunas, the distance is still 364 km, much more than 310.

Considering the very close match between the road distance and the distance used by Z/F,

What close match is that? In 3 out of 4 cases, there are very considerable differences.

I have a hard time believing that what they used is aerial distance in all cases.

Why is that, if, as you claim, road distances and ACTF are pretty much the same anyway? We know exactly what they used, as this is clarified in a footnote - they used the distances given in the Kartenband to HGPW.

[Qvist]

I'm out of this. I have better uses for my adrenalin.

[Andreas]

You have given me three examples. One of these is buggy, and the correction you posted is the absolute upper bound, in fact I think it will be about half that added distance, but I check the actual jumping off point of 8.PD in Leeb and Beaulieu tonight. The second one shows 20% difference. The third one shows 2% difference. Those are not very considerable, in fact in the second case I'd put it down to differences in calculation. Only in the case of Pliyev I am to believe that we have 43% difference between road and aerial distance.

Sorry, I don't believe that for a minute, I stick to my conclusion that there is an error. I also don't understand what the problem with that conclusion is. Errors happen all the time, I am responsible for a book that we just published where we had to put out an errata sheet almost immediately, due to an embarrassing error. Shrug - that's life.

All the best

Andreas

[Andreas]

I'm out of this. I have better uses for my adrenalin.

I am sorry to hear this, but I do not understand what you are getting so upset about.

All the best

Andreas

[Qvist]

You have given me three examples. One of these is buggy, and the correction you posted is the absolute upper bound, in fact I think it will be about half that added distance, but I check the actual jumping off point of 8.PD in Leeb and Beaulieu tonight. The second one shows 20% difference. The third one shows 2% difference. Those are not very considerable, in fact in the second case I'd put it down to differences in calculation. Only in the case of Pliyev I am to believe that we have 43% difference between road and aerial distance.

I have given you four examples - the three rates for Barbarossa, plus Pliyev. I had half a mind to sit down and produce more, but that's when i decided I have better uses for my adrenalin.

Sorry, I don't believe that for a minute, I stick to my conclusion that there is an error. I also don't understand what the problem with that conclusion is.

The problem is that you are assuming it without any clear proof, and rather than find out what sort of distance measurement has been used, what dates, or along what sort of axis, you assume that these are wrong. You make a big stink out of using the 23rd as a starting date, without bothering to find out if the logic and purpose of the calculations makes this a viable choice, which they do. The headings even say "5 GTA & 11GA", for crying out loud.

It is of course possible that there are mistakes in the article, I am just not prepared to assume it without proof. If I were you, I'd rather look at the timelines for the Pliyev group, which do strike me as rather odd, since the 1st advance ends on the 8th and the second begins on the 4th. But there is of course no way of knowing why, without further investigations.

Errors happen all the time, I am responsible for a book that we just published where we had to put out an errata sheet almost immediately, due to an embarrassing error. Shrug - that's life.

No big deal then? No reason to get into characterisations like "appears very fishy to me", "inclined to reject Zetterling/Frankson's whole analysis as ill-researched and unverifiable based on the sources, and I am left wondering what kind of peer review is happening at the journal they published in" or "I am extremely unimpressed by the quality of Zetterling/Frankson's analysis." ? Without having read the article first?

In short - I do not enjoy having to play the sword-bearer of an article I haven't written just because who critiques it goes way over-board in his approach, and probably I am only making it worse by doing so. Hence, I choose to bow out.

And in any case it does not even matter much to the basic point - you can upgrade Pliyev's rate to 30 km/d, and it does not change the overall picture significantly.

cheers

[Andreas]

If I were you, I'd rather look at the timelines for the Pliyev group, which do strike me as rather odd, since the 1st advance ends on the 8th and the second begins on the 4th. But there is of course no way of knowing why, without further investigations.

I have looked at the timeline for Pliyev's group, and the date of 8th does not make a lot of sense to me in terms of how the Soviets seem to have grouped the phases. The first advance does not end on the 8th. It ends on the 29th, before the battle of Slutsk, with the destruction of the Bobruisk pocket. The second phase ends on the 4th, with the Minsk encirclement. The third phase ends on the 16th July. For Pliyev's group the three stages are participation in the encirclement of Bobruisk, taking Slutsk, assisting in taking Baranovichi, which fell on July 8th.

All the best

Andreas

[Andreas]

You have given me three examples. One of these is buggy, and the correction you posted is the absolute upper bound, in fact I think it will be about half that added distance, but I check the actual jumping off point of 8.PD in Leeb and Beaulieu tonight. The second one shows 20% difference. The third one shows 2% difference. Those are not very considerable, in fact in the second case I'd put it down to differences in calculation. Only in the case of Pliyev I am to believe that we have 43% difference between road and aerial distance.

I have given you four examples - the three rates for Barbarossa, plus Pliyev. I had half a mind to sit down and produce more, but that's when i decided I have better uses for my adrenalin.

Yes, and the only one that is significantly out is Pliyev.'s first stage Which is my point.

Sorry, I don't believe that for a minute, I stick to my conclusion that there is an error. I also don't understand what the problem with that conclusion is.

The problem is that you are assuming it without any clear proof, and rather than find out what sort of distance measurement has been used, what dates, or along what sort of axis, you assume that these are wrong.

For the next phase (8 July to 23rd July), we again have an almost complete match on Viamichelin (202km) and Z/F (197km). It should also be clear that the time between 23rd June and 27 July does not add up to 38 days (23 + 15) as is claimed in Z/F. The total distance from Svetlahorsk to Bialystok is 518km (road follows the major axis of advance), not 426km.

What sort of proof do you need to accept that the 229km is just plain wrong, as is either the end of of the first phase in Z/F at the historically correct date of the Stavka order to exploit west (which had no relevance for the CMG), or the beginning of the second advance on the historically wrong date in Z/F (which mattered for the CMG because Baranovichi fell on the day)?

You make a big stink out of using the 23rd as a starting date, without bothering to find out if the logic and purpose of the calculations makes this a viable choice, which they do. The headings even say "5 GTA & 11GA", for crying out loud.

We are coming back to apples and oranges again Qvist. If you think that comparing an all-arms army advance at the end of which this army is involved in a large-scale encirclement battle (11 GTA) to that of a fully motorised spearhead forming a small bridgehead without comparable combat activity (LVI.AK) is a viable choice, you are welcome to do so. I don't. I see this as an attempt to simplify the analysis (which is not a bad intention), but one that raises far more questions than it answers.

Errors happen all the time, I am responsible for a book that we just published where we had to put out an errata sheet almost immediately, due to an embarrassing error. Shrug - that's life.

No big deal then? No reason to get into characterisations like "appears very fishy to me", "inclined to reject Zetterling/Frankson's whole analysis as ill-researched and unverifiable based on the sources, and I am left wondering what kind of peer review is happening at the journal they published in" or "I am extremely unimpressed by the quality of Zetterling/Frankson's analysis." ? Without having read the article first?

I presumed that what you quoted initially was correct as far as the article goes, and that was what these remarks were based on. I note that you made an error about the 3rd in the tables you typed. I stick to the 229km by 8th July being wrong in the article, I find none of your explanations convincing. I also stick to aerial distance being meaningless if you want to compare advance rates. There is also what appears to be an error in the major axis of the CMG, where the drive north of Stolbtsy is excluded. I am also unhappy about the inclusion of intermediary objectives for a total failure of an operation (Typhoon 1st phase).

And in any case it does not even matter much to the basic point - you can upgrade Pliyev's rate to 30 km/d, and it does not change the overall picture significantly.

Quite, but I think we both would arrive at the same conclusion for vastly different reasons. 30km/d (road, not aerial) is pretty close to what the northern CMG achieved in Bagration by July 3rd, by the way.

All the best

Andreas

[Qvist]

I can only suggest that you contact Niklas Zetterling to discuss this - I haven't written the article, I don't know anything more than you do about how the calculations were done or why, and there might be perfectly good explanations on each of these points that do not occur to either of us.

Anyway, this effectively leaves us without a single advance rate on whose viability there's any consensus, so unless someone is prepared and able to produce ones that meet with general approval, we appear to have no further basis for the discussion.

cheers

[Andreas]

As I pointed out above, advance rates are not really that useful in my view, since they are a function of multiple inter-related elements. I'd be much more interested in picking two operations that are similar in as many aspects as possible for comparison, and try to refine Jon's model on them. Similarities could/should be force size, depth of total advance, outcome, duration, enemy losses inflicted as share of total enemy force, etc.

I think that would be a far more interesting discussion.

All the best

Andreas

[TRose]

If you want to see how fast a non motorized force can advance against active oppsition, go look at the Chinese initial attack in Korea. forget how fast they where able to advance, but it was also done in the face of the UN forces having complete air superiority.

[Qvist]

I'd be much more interested in picking two operations that are similar in as many aspects as possible for comparison, and try to refine Jon's model on them. Similarities could/should be force size, depth of total advance, outcome, duration, enemy losses inflicted as share of total enemy force, etc.

I think however that it will be difficult to impossible to find two operations (one soviet, one German) who are fairly similar in all these respects, or in almost all.

The closest possibility I can think of is HG Mitte's Minsk and Bialystok pocket battle in June/July, and Iassy-Kishinev. These had pretty similar force sizes for the attacker (around 1.3 million). I suppose we can tailor the HG Mitte advance to fit Iassy-Kishinev, and put it at nine days, or at least close to that - the Minsk Kessel was closed on 28-29 June. The defender in both cases lost a large percentage of the forces committed, and both battles can be described as having resulted in the annihilation of the defending forces.

But the depth of advance was considerably greater for HG Mitte (442 km in 7 days, compared to 320 km in 9 days), the defending force was numerically considerably stronger than at Iassy-Kishinev, and the losses suffered by the attacker was less than a third of what it was at Iassy-Kishinev ( 67,130 at Iassy-Kishinev, roughly 23,000 to 6 July (or 14 days rather than 9) for HG Mitte). Also, there were special circumstances with a large degree of impact in both cases - the strategic surprise of Barbarossa in one, the defection of Romania in the other. Still, they are as close as anything I can think of, at least.

Another important variable that should be similar in both cases: Time of year.

Another thing - if we compare operations with similar variables using a relatively rough model such as proposed, they will inevitably yield similar results. The point of the model is to enable comparing operations that are not similar in the variables, no?

cheers

[natham]

german advance into france and russian operation bargation of army group center.in this i mean total displacment of oppositions defence against an aggresive opponent and there for breakthrough on a massive scale.

[Qvist]

These were however

a) between different opponents and in different theaters of war and
b) Massively dissimilar in many of the variables.

cheers

[JonS]

Another thing - if we compare operations with similar variables using a relatively rough model such as proposed, they will inevitably yield similar results.

Not necessarily, but even if they do that tells you something. See eg Andreas' comparison of AGN in Barbarossa and Bagration - they scored similarly overall, but got there by different means. Don't you think that is an interesting result? I think it is, since it implies that the overall sophistication of German logistics in AGN Barbarossa were on a par with Bagration, despite the differences.

The point of the model is to enable comparing operations that are not similar in the variables, no?

Yes, of course. That's why there are six (or seven - success/failure [which I'm not too sure is necessary] ... or eight [weather/time of year/climate]) factors. With three levels for each factor, that gives 729 different 'types' of offensive (2187 with seven factors, 6561 with eight factors)

Yes, it's rough. Partly that's because I haven't put much time into it yet, but mainly it's deliberate so that we can get away from pointless quibbling about whether the average daily rate of advance was 41.8km or 43.2km, or whether the advance started on the 2nd of October at 2pm, or on the 3rd at 5am. Etc. If you wanted to be a bit more rigorous you could take a group of, oh I don't know, say a dozen offensives you think are representative and rank them from first to last on each of the factors (Bagration had more forces than AGN, but AGN travelled further overall, etc). Then the top third in each factor get HIGH, the middle third get MED, and so on. Then total each offensive for it's own factors, then compare the overall offensives to each other. That would tell you which offensives had 'good' log spt, and which had 'poor' log spt.

Edit after a-run-and-a-think: It would also allow you to do useful comparisons within and between countries. Eg "oh, look - the Germans consistently get a higher average advance, but the Russians consistently advance further overall" or "oh look, the British consistently score LOW for forces engaged", etc.

In short, it's a way of removing the oranges from comparisons.

[Andreas]

To add to what Jon just wrote, with which I agree, I would like to clarify that 'similarity' does not mean 'identical'. So e.g. I would see no need to normalise duration. Similar operational lengths could be (say, in a range from 20-25 days, or whatever, but e.g. 7 and 18 days would not be similar enough). Similar forces would be in very broad range again. Etc. Similarity is just a very rough requirement.

All the best

Andreas