Japanese Versus American Victory Claims (6 Viewers)

[Frog]

Japanese bomber and attack plane losses were often pretty appalling even in the first year of the war, especially during attacks on warships.

The Japanese concluded they needed something like a 3:1 to 5:1 ratio of fighter cover to bombers in order to execute daylight strikes in contested airspace. Lots of Japanese strike packages over the Solomons and New Guinea in 1943 featured 10 or fewer bombers plus dozens of fighters split between close cover duty and enemy fighter-hunting. Even still the losses were sometimes unsustainably high.

I'd attribute a lot of the extremely high loss rate to allied anti-aircraft capabilities and not just air superiority. Attacking an American task force by the last year of the war meant flying into radar directed guns firing proximity fuze rounds and literally hundreds of 40mm Bofors and 20mm Oerlikon. It took a lot of training and skill to pull off accurate torpedo and dive bombing attacks even in more or less ideal conditions, let alone under that kind of fire. So not only did the losses become completely unsustainable, the results were fewer and fewer. By Formosa in October, 1944, you had some sizeable formations of Japanese attack planes and bombers achieving tactical surprise by approaching US carriers at low level below radar, and nevertheless incurring crippling and irreplaceable losses with only a handful of bomb and torpedo hits to show for it.

It figures kamikaze was an anti-ship tactic.

Not even the last year of the war :during the Santa Cruz Island battle (25–27 October 1942), even if tactically a japanese navy victory, they had to cope with an air defense they never saw before resulting with an appaling cost for the Kido Butai, in term of trained crewmen and pilots and above all of an invaluable veteran cadre loss (groups, squadrons and sections leaders).

 

[andrewgeorge]

Not even the last year of the war :during the Santa Cruz Island battle (25–27 October 1942), even if tactically a japanese navy victory, they had to cope with an air defense they never saw before resulting with an appaling cost for the Kido Butai, in term of trained crewmen and pilots and above all of an invaluable veteran cadre loss (groups, squadrons and sections leaders).

That's what I had in mind when I said losses were sometimes appalling even in the first year of the war. Actions like Santa Cruz or the attack on Lexington off Rabaul even earlier in the war. At least at Santa Cruz the attack and bomber pilots dished it out while getting mauled.

I'm not even sure I'd call Santa Cruz really a victory—I mean, I guess strictly speaking the American fleet lost more ships and retired from the engagement zone. But it was in support of an offensive on Guadalcanal that failed horribly, so the Japanese didn't end up really accomplishing any operational objective.

 

[Frog]

That's what I had in mind when I said losses were sometimes appalling even in the first year of the war. Actions like Santa Cruz or the attack on Lexington off Rabaul even earlier in the war. At least at Santa Cruz the attack and bomber pilots dished it out while getting mauled.

I'm not even sure I'd call Santa Cruz really a victory—I mean, I guess strictly speaking the American fleet lost more ships and retired from the engagement zone. But it was in support of an offensive on Guadalcanal that failed horribly, so the Japanese didn't end up really accomplishing any operational objective.

At least a tactical victory : the damaged Enterprise had to have been jury rigged and for a little span of time, the only flat top (temporarily) on stage was japanese.
But the dire outcome was to come later.
To sum up :
- the core of carriers were lost at Midway ;
- the core of pilots/crewmen was lost at Santa Cruz.

 

[andrewgeorge]

At least a tactical victory : the damaged Enterprise had to have been jury rigged and for a little span of time, the only flat top (temporarily) on stage was japanese.
But the dire outcome was to come later.
To sum up :
- the core of carriers were lost at Midway ;
- the core of pilots/crewmen was lost at Santa Cruz.

That's probably a fair shorthand. Attrition kept Japanese carrier air groups from ever really fully reconstituting their skilled aircrews over 1943-1944 as well. There weren't many shipboard combat deployments since the allies were largely on the offensive. And yet, every time additional air power was needed in the Southeast Area (every few months), a carrier division's air groups would deploy temporarily down to the South Pacific from Truk, where they would incur additional losses which prolonged the rebuilding.

By the time you got to Philippine Sea, most carrier aircrew simply didn't have the training and experience to be very combat effective, and so the US Navy (with all of its other material advantages—numbers of aircraft, numbers of ships, radar-directed intercept, radar fire control, etc) simply chewed them up.

 

[GregP]

So much about aviation is covered incorrectly in the media that AI tends to retrieve the incorrect stuff as gospel. I've spent a lifetime around aircraft / aviation. Hardly any media civerage about aviation is anywhere near true. Quite a few of the books used as references by AI are nothing more than collections of stories with the attendant embellishment that always creeps in when stories are told.

When I have used AI to get things about WWII aviation, it is about 40-60% accurate if I'm lucky. Many times I'm not so lucky. AI doesn't sift through text for accuracy; it sifts for published contant regardless of accuracy. It basically outputs what it finds written in text that is avaiable for perusal.

It's kind of like a self-driving car. The "goodness" of the AI driver is completely dependent upon the programmer or team that writes the software, and they're only as good as the things they anticipate might happen in traffic. If anything happens they have not anticipated, the result is a toss up as to whether or not the response is good. Witness the stupid things done by self-driving taxis fairly recently.

 

[Squish]

So much about aviation is covered incorrectly in the media that AI tends to retrieve the incorrect stuff as gospel. I've spent a lifetime around aircraft / aviation. Hardly any media civerage about aviation is anywhere near true. Quite a few of the books used as references by AI are nothing more than collections of stories with the attendant embellishment that always creeps in when stories are told.

When I have used AI to get things about WWII aviation, it is about 40-60% accurate if I'm lucky. Many times I'm not so lucky. AI doesn't sift through text for accuracy; it sifts for published contant regardless of accuracy. It basically outputs what it finds written in text that is avaiable for perusal.

It's kind of like a self-driving car. The "goodness" of the AI driver is completely dependent upon the programmer or team that writes the software, and they're only as good as the things they anticipate might happen in traffic. If anything happens they have not anticipated, the result is a toss up as to whether or not the response is good. Witness the stupid things done by self-driving taxis fairly recently.

A large factor in AI-supported arguments is that the LLM is usually coded to also agree with whoever is writing the prompt and doesn't really have any standard for intellectual consistency either. Also note how the specific anecdote that was started with was relating to a completely different author by the name of Ichimuru.

I should admit that I am no expert on the history of the Central Pacific around 1943, but I can offer some examples from Umemoto's Kaigun Zero-sentai Gekitsui Sentōki where he mostly counting losses & claims from 201st, 204th and 253rd Kokutai. (Umemoto's statistics may have influenced Claringbould's narrative, as his work published after Umemoto's.) A good example would be, on July 2, 1943, a joint operation was carried out to bomb Allied installations. Umemoto described the result as one Allied Corsair lost and two more forced to land, with no Japanese losses, and concluded that the exchange was "1:0." However, IJAAF Ki-61s and Ki-43s also took part, and at least three Ki-61s failed to return. These losses were not included in Umemoto's account.

The criticism here seems to be that Ichimuru was taking Japanese claims at face value and also discounting the losses of the IJAAF. So where do his figures for IJAAF losses come from in his rebuttal? We do not know; for all we know he has just inserted US claims at face value?

However what is more telling is when you actually look at reports from July 2 1943 from the units that he has named.

There is no online record from 02 July 1943 for the 201st Kokutai available online. This is the first report available from 15 July 1943. The same can be said of the 253rd Kokutai. The other two units that are listed in the criticism did not participate in the attack.
The report from the 204th Kokutai is available and we can see that the 204th claimed 4 aircraft destroyed. This is already 4x that of what is allegedly cited in Ichimuru / Umemoto's book, and only represents one of the allegedly 3 participating IJNAF fighter units.

The fact of the matter is that the AI program he used probably translated Umemoto's book for him, and then completely messed up the units involved.


This is the report from the 251st Kokutai from the same day. There is some kind of claim in the document but the text is extremely faded / illegible. Supposedly it says that is 2 F4F's claimed with one of those being probable; even if we don't accept that...we can see that there is some form of claim in the box that is used for claims.

And here is the last relevant Japanese record from that day. This is from the 582nd Kokutai which also furnished escort for the July 2nd raid. In this report we can see claims for 4 fighters shot down.

The IJNAF claims from this raid amount to 10 US Fighters shot down. If Ichimuru / Umemoto was reporting Japanese claims at face value then his book would say that Japan had a 10:0 loss ratio on 2 July 1943, and not the 1:0 that the OP alleges. This is for an air raid in which IJAAF provided the bombers, and the IJNAF provided the fighter escort.

There is a raid on 4 July 1943 that largely follows the same formula, but was furnished escort by both IJAAF and IJNAF and in which they did take losses. His AI program has probably conflated both raids together and hallucinated events from 04 July to be taking place on 02 July in order to refute an imagined argument.

 

[Irregualr23]

So much about aviation is covered incorrectly in the media that AI tends to retrieve the incorrect stuff as gospel. I've spent a lifetime around aircraft / aviation. Hardly any media civerage about aviation is anywhere near true. Quite a few of the books used as references by AI are nothing more than collections of stories with the attendant embellishment that always creeps in when stories are told.

When I have used AI to get things about WWII aviation, it is about 40-60% accurate if I'm lucky. Many times I'm not so lucky. AI doesn't sift through text for accuracy; it sifts for published contant regardless of accuracy. It basically outputs what it finds written in text that is avaiable for perusal.

It's kind of like a self-driving car. The "goodness" of the AI driver is completely dependent upon the programmer or team that writes the software, and they're only as good as the things they anticipate might happen in traffic. If anything happens they have not anticipated, the result is a toss up as to whether or not the response is good. Witness the stupid things done by self-driving taxis fairly recently.

Free AI tools simply tend to make up stuff, using sota model and anything above gpt5.6 and fable5 mostly get rid off the hallucination you see in older models, and they are very useful for going over large quantities of images and ww2 orals in batches and always linking evidence with original source rather than make up any fact by their own.

AI is definitely doing better on reading and understanding context than those who doesn't understand what researcher's writing based on losses reported on each side rather than pure claims.

 

[Squish]

From the USSBS post-war interrogation and reconstruction, neither those 64 or 44 losses held valid for the Japanese. The true loss number to all causes was around 120 - 130 for Jan 1944, for IJNAF only.
View attachment 898976

Free AI tools simply tend to make up stuff, using sota model and anything above gpt5.6 and fable5 mostly get rid off the hallucination you see in older models, and they are very useful for going over large quantities of images and ww2 orals in batches and always linking evidence with original source rather than make up any fact by their own.

AI is definitely doing better on reading and understanding context than those who doesn't understand what researcher's writing based on losses reported on each side rather than pure claims.

The chart that is quoted here as the source showing 120-130 losses for the IJNAF in January 1944 is actually just showing the amount of planes that they had on hand in the Solomon Island campaign. The AI research assistant did not read the chart correctly at all. It is actually just a slightly different version of the immediately proceeding chart.

Note how it shows the 11th Air Fleet having around 125 aircraft on hand in January 1944. The chart he has cited has nothing to do with documenting monthly IJNAF monthly losses.

The major difference is that this chart focuses more exclusively on the Solomon Islands and starts in August of 1942.

This chart here is made using the actual indexed numbers in the US SBS interrogation. Japanese fighters lost due to combat in January 1944 are listed as 80, with another 183 being classified as non-combat losses. In the US SBS basically any plane that is destroyed on the ground was considered a non-combat loss, although this number might also include planes that were damaged airborne and returned to base to be written off.

 

[DerAdlerIstGelandet]

S Squish I Irregualr23

Please take your argument elsewhere. The bickering and personal attacks you are making toward each other has a negative impact on everyone's forum experience.

You can both debate this topic in a civil manner without the sideshow.