ASW Stuff

[Local Yokel]

Fine, but that leaves unanswered the question of what historical capability is being achieved. The Japanese C- and D-gata kaibokan had the historical capability of firing 12 d/c launchers on a single run. Their game equivalents only have a capability of firing 6. That is somewhat counter-intuitive.

[Sardaukar]

I don't think you really read what is said....

Just mod it and give them historical numbers...and see how code treats it.

[Local Yokel]

ORIGINAL: Sardaukar

I don't think you really read what is said....

What do you believe I have missed?

ORIGINAL: Sardaukar

Just mod it and give them historical numbers...and see how code treats it.

Sorry, don't understand. All I will see will be a set of combat results, but I will be none the wiser as the way in which they were affected by code to which I have no access. That's why I am hoping that someone who does enjoy such access can explain why they don't consider the code handles the historical numbers well.

[vettim89]

Sorry, don't understand. All I will see will be a set of combat results, but I will be none the wiser as the way in which they were affected by code to which I have no access. That's why I am hoping that someone who does enjoy such access can explain why they don't consider the code handles the historical numbers well.

Well thats the rub now isn't it? The powers that be have, in my opinion, correctly decided that the intricacies of how the code works are not for public distribution. As modders that puts us into the position of having to use trial and error to discover how our changes to the database affect the results

To clarify, I was not intending to broaden this into a discussion into a general discussion about Japanese ASW capabilities. The facts I stated above were merely to add substance to the case for looking at this issue. Put a "X" down for me as one who greatly applauds the DaBabes team willingness to at least look at it.

[morganbj]

ORIGINAL: JWE
Historical accuracy is a model, a goal, a touchstone. Where it can be achieved, well and good. Where not, something must be devised that gives acceptable results but maintains consistency.

This is what I've been saying for years about AE. Its individual systems are quite good, but imperfect, but when taken together, they render a perfectly acceptable model of the strategic war. Quite frankly, AE works.

I, too, still have my pet peeves about some of the systems (e.g., land combat), but I've learned to adapt my play to account for those things that seem a tad off. I may, for instance, apply a few extra units to an attack to make sure I have some success, or, I may modify how and where I use my submarines to avoid getting massacred by IJN Es. But the war still seems to go according to the capabilities of the participants. There are more than ample strategic possibilites, and each plays out a little differently, but probably close to what would actually have happened had that strategy been tried. I find it remarkable that it all works so well. Yes, I'd like a perfect program, but AE is good enough for me. I can certainly understand the feeling that a few vocal people can cause additional unanticipated problems when the code is changed to make something "work better." (Remember the artillery fix, then refix, unfix, and fix again? And it's still not quite right.) Generally, though the code is in very good shape; most of the serious concerns have indeed been addressed. And, it'll get better.

That said, I also greatly appreciate what the Da Babes team have done. I realize Da Babes doesn't deal with the code, but makes modifications to the data files to accomplish its ends. That means that Da Babes is forced to use an imperfect engine, one that is in pretty good shape, but not without a few issues that may never be fixed. After all, it's impossible to code reality perfectly. Da Babes makes that imperfect code seem to work a little less imperfectly. That's called "success."

So, keep up the good work. I'll play whatever you guys conjure up. I trust your decisions. I know you're thinking things through. Besides, like you say, if I don't like what you do, I can use the editor, too.

Keep pluggin' away.

[treespider]

My take on what Local Yokel was trying to say :

Do we have a historical baseline on which to judge the E's?

In other words perhaps the E's in game performance is being "enhanced" by other factors, such as operational employment of the hunters and hunted, air search assets etc. that are different than history.

Personally I have no issue with a one-off "correction" if the system is an "Outlier" and the "correction" brings the system into line with our perceived reality.

[vettim89]

ORIGINAL: treespider

My take on what Local Yokel was trying to say :

Do we have a historical baseline on which to judge the E's?

In other words perhaps the E's in game performance is being "enhanced" by other factors, such as operational employment of the hunters and hunted, air search assets etc. that are different than history.

Personally I have no issue with a one-off "correction" if the system is an "Outlier" and the "correction" brings the system into line with our perceived reality.

I have been contemplatingthis for a day now and I find myself embracing this idea more and more. Watching combat replays may give us an idea of what the code is doing during ASW resolution. Comments include

"Sub evades pattern"
"USS XXX unable to locate submarine"
"USS XXX unable to get position to attacke submarine"
"Escorts searching for submarine"
"Near miss rattles submarine"

So perhaps its not just about weapons moints and how they are displayed in the DB. Maybe ASW rating, which of course is related to the number of weapons mounts, is just as big a factor. If true then the "E" class patrol craft will still have the obnoxious ASW of 12 no matter how the mounts are grouped. Just throwing that out there for consideration

[JWE]

ORIGINAL: vettim89
So perhaps its not just about weapons moints and how they are displayed in the DB. Maybe ASW rating, which of course is related to the number of weapons mounts, is just as big a factor. If true then the "E" class patrol craft will still have the obnoxious ASW of 12 no matter how the mounts are grouped. Just throwing that out there for consideration

nooo ... ASW rating has no impact. It is nothing but a WeapNum summation shown on the ship screen that people can use to roughly judge who's better then who. There are 3 different things that happen; in sequence, but with some overlap.

1) Acquisition - You need a good Detect value to find and hold a sub for prosecution. Leader and crew exp values are of primary importance.
2) Prosecution - Done on an individual ship-by-ship basis. Crew exp values are of primary importance.
...2a) How many weapons (i.e., slots) will engage? Crew exp values are of primary importance.
...2b) How many passes (i.e., attacks) will the ship make? Crew exp values are of primary importance.
...2c) Before each pass, will the ship acquire/reacquire the sub? Crew exp values are of primary importance.
3) Destruction - Done on an individual WeapSlot-by-slot basis. WeapNum is of primary importance.
...this is the only place in the wide, wide, world of sports, where WeapNum of a slot is important. This is the % chance to hit part. All the rest is just % chance to acquire and % chance to shoot.

So all of your 'messages' relate to item ...2c), except the last one which indicates 'close but no cigar' under item 3). Before you drop, the sub can evade. You can lose contact while moving into position to drop. You might just be too far away to effectively prosecute. None of these things have anything to do with how many DC launchers (or slots) you gots. That happens in part 3). And that's where the math comes into play.

This is NOT THE ALGORITHM. It is an UTTER SIMPLIFICATION, but it might give you some idea of what's involved. A slot might have Num of 2, 3, 4. Think of 2^Num and you get 4, 8, 16. Think of that as a % chance to hit with that slot. Now make a slot with Num of 12 or 16 (the original Es). 2^Num is 4096 or 65536. A little bigger than 16, yes? So if an original E could acquire, it could prosecute (those 12-16 launchers were in 1 slot). If it could shoot it WOULD KILL. That was the booger. And a right wet and green one it was, too.

I really hope this makes sense to ya'll. Ciao. John

[oldman45]

In fact that does make sense. It now brings another question which might be better served in another thread.

Since crew exp plays such a large role, how then is the best way to get the US "escort" crews up to speed so that in 1943/44 they do a great job of keeping the subs at bay? In my game I try to always use the same group of "escorts" doing nothing but ASW hunting but I do not see a lot of change in exp.

[vettim89]

ORIGINAL: oldman45

In fact that does make sense. It now brings another question which might be better served in another thread.

Since crew exp plays such a large role, how then is the best way to get the US "escort" crews up to speed so that in 1943/44 they do a great job of keeping the subs at bay? In my game I try to always use the same group of "escorts" doing nothing but ASW hunting but I do not see a lot of change in exp.

Just steaming will raise exp but it has a cap. I beleive it has been put forth that it is 55. Beyond that the only way to get exp to increase is actually being involve in combat

[vettim89]

ORIGINAL: JWE

ORIGINAL: vettim89
So perhaps its not just about weapons moints and how they are displayed in the DB. Maybe ASW rating, which of course is related to the number of weapons mounts, is just as big a factor. If true then the "E" class patrol craft will still have the obnoxious ASW of 12 no matter how the mounts are grouped. Just throwing that out there for consideration

nooo ... ASW rating has no impact. It is nothing but a WeapNum summation shown on the ship screen that people can use to roughly judge who's better then who. There are 3 different things that happen; in sequence, but with some overlap.

1) Acquisition - You need a good Detect value to find and hold a sub for prosecution. Leader and crew exp values are of primary importance.
2) Prosecution - Done on an individual ship-by-ship basis. Crew exp values are of primary importance.
...2a) How many weapons (i.e., slots) will engage? Crew exp values are of primary importance.
...2b) How many passes (i.e., attacks) will the ship make? Crew exp values are of primary importance.
...2c) Before each pass, will the ship acquire/reacquire the sub? Crew exp values are of primary importance.
3) Destruction - Done on an individual WeapSlot-by-slot basis. WeapNum is of primary importance.
...this is the only place in the wide, wide, world of sports, where WeapNum of a slot is important. This is the % chance to hit part. All the rest is just % chance to acquire and % chance to shoot.

So all of your 'messages' relate to item ...2c), except the last one which indicates 'close but no cigar' under item 3). Before you drop, the sub can evade. You can lose contact while moving into position to drop. You might just be too far away to effectively prosecute. None of these things have anything to do with how many DC launchers (or slots) you gots. That happens in part 3). And that's where the math comes into play.

This is NOT THE ALGORITHM. It is an UTTER SIMPLIFICATION, but it might give you some idea of what's involved. A slot might have Num of 2, 3, 4. Think of 2^Num and you get 4, 8, 16. Think of that as a % chance to hit with that slot. Now make a slot with Num of 12 or 16 (the original Es). 2^Num is 4096 or 65536. A little bigger than 16, yes? So if an original E could acquire, it could prosecute (those 12-16 launchers were in 1 slot). If it could shoot it WOULD KILL. That was the booger. And a right wet and green one it was, too.

I really hope this makes sense to ya'll. Ciao. John

John,

That does make sense. However, when I ran my naked ASW tests last year I found that ASW value does play a big role in a TF's ability to detect a sub that starts the turn with a detection level of 0. I ran a test bed scenario where all other factors were the same (same ships, same crew exp, same environment). What I changed was the ASW value. I stared with 2 then went to 3 then 6. With each succeeding increase I saw a higher number of detections then prosecutions by the ASW TF. The test had the ASW TF's deliberately passing over the top of undetected subs. So while the ASW value does not have an effect on the combat once it starts, it does seem to have an affect on if there will be combat.

[witpqs]

ORIGINAL: JWE

...2a) How many weapons (i.e., slots) will engage? Crew exp values are of primary importance.

This one is a very, very nice detail. I shortly ago finished reading Turning The Tide, which focused on a few convoys at that point in the Battle of the Atlantic during which things went from bad to good for the Allies. There were many instances cited of an escort dropping only one (1) depth charge!

I know that earlier someone posted that such was never done and only patterns used. While that isn't strictly true, the post wasn't totally off either. What I mean is that the "1" depth charge attacks were used when they did not have a very good contact on the u-boat, with the intention to keep it down and hiding instead of trying to maneuver back into attack position. They used full patterns when they had decent contact and thought they could damage or sink the u-boat. They also fired/dropped between 1 and "full" (which varied by ship of course), just depending on the captain's judgement.

The number of depth charges on board and how much longer the convoy had to run versus when they could make port to rearm were key considerations.

So, this line item about how many weapons will engage is a quite nice addition to the realism.

[JWE]

ORIGINAL: oldman45
Since crew exp plays such a large role, how then is the best way to get the US "escort" crews up to speed so that in 1943/44 they do a great job of keeping the subs at bay? In my game I try to always use the same group of "escorts" doing nothing but ASW hunting but I do not see a lot of change in exp.

As vettim says, "Just steaming will raise exp but it has a cap." Ok, true. But just 'hunting' won't do either. One must kill in order to know how to do so better.

Local Yokel has an excellent IJN "training" approach that might be of benefit to other nationalities. There's some code things that happen annually, from '42 on, that deal with some of this stuff, but specific and directed training, a la LY's method, is a good thing all around.

[Local Yokel]

ORIGINAL: JWE

This is NOT THE ALGORITHM. It is an UTTER SIMPLIFICATION, but it might give you some idea of what's involved. A slot might have Num of 2, 3, 4. Think of 2^Num and you get 4, 8, 16. Think of that as a % chance to hit with that slot. Now make a slot with Num of 12 or 16 (the original Es). 2^Num is 4096 or 65536. A little bigger than 16, yes? So if an original E could acquire, it could prosecute (those 12-16 launchers were in 1 slot). If it could shoot it WOULD KILL. That was the booger. And a right wet and green one it was, too.

That is a lucid explanation of the consequences of having a large number of 'tubes' in a single weapon slot. It makes it entirely comprehensible that an IJN Escort that fires a 12- or 14-tube weapon slot is more or less assured of a kill.

It seems to me that it also makes it clear that you have little to fear by restoring to those E's their full complement of 12 or 16 side throwers provided you break that complement down into multiple slots. Adopting JWE's example, if a C- or D-gata kaibokan's 12 throwers are broken down into four groups of 3 (2 on each side) as suggested in my post #13 then each slot (if it fires at all - JWE's 2a) has a 2^3 % chance of a hit = 8%.

Compare and contrast with the Tacoma/Colony class DE's with their single bank of 8 centreline Y-guns (this is actually wrong – the Tacomas mounted 4 K-guns on each side, so the weapons fit should be reconfigured for this reason alone). If this 8-tube slot gets to fire then it's 2^8 % chance of a hit = 512% - bye,bye I-boat. This is why I think this class requires further attention.

Treespider was absolutely on the mark: I was concerned that halving the number of throwers on the E’s might have been prompted by a mistaken perception of what was required for historical accuracy. Not only might these ships' apparently overpowered performance have been the product of game circumstances (e.g. better air search raising DL's, better tactics by Player 2), but also our perception of history might be adrift. Not too long ago, thanks to Fuchida Mitsuo, conventional wisdom had it that at Midway Enterprise and Yorktown’s SBDs caught KdB with flight decks crammed with strike aircraft at the point of launch. Now we have good reason to think otherwise – our perception of history was fundamentally flawed. The same might have been the case as regards our perception of Japanese ASW ships' capabilities.

In the event, I think JWE's clarification shows that the only problem with the supposedly overpowered E's is the excessive concentration of tubes in a single slot, and the good news is that it seems that these ships can still carry their full complement of throwers without becoming excessively powerful by parcelling those throwers out into multiple slots. In other words the standard method of modelling a ship's armament still holds good.

On training and experience levels, I strongly recommend people to take a look at section 3.3.1.1 of the Editor manual, which is highly instructive. What this shows is that in 1941-42 non-British Allied crews arrive at 80% of the base experience level (Note 1), and from 1943 onwards at base level + random 0-10. Japanese crews enjoy an experience advantage in the first two years, but thereafter the difference between opposing crews' initial experience is minimal. Note that:

a) Allied DEs' base experience level is 50. For E-class ships it is unspecified but appears to be 40.
b) The Japanese are subject to a significant experience disadvantage in ASW due to their negative modifier and the Allied crews' positive modifier. Getting decent ASW performance out of Japanese crews will be no easy matter.

<edit> Note 1: plus random 0-10 </edit>

[oldman45]

Thanks John, Thanks LY.

[el cid again]

At least we are getting some information about how things work in code. It probably isn't the way things ought to work,
but unless someone decides to change it, it is what we have. The problem here is that other weapons tend to shoot only
if there is a target out there to shoot at - on their bearing. ASW weapons tend to shoot in patterns, somewhat blindly,
to make up for the lack of precision of target location (the target moves during the fall of shot, as it were, and there is
usually time between detection and reaching the drop point as well - during which the target may maneuver - and it almost
certainly changes position even if not course). So ideally we should consume shot for all the weapons, even if only one
is able to "bear" - as it were. That still sounds to me like you want to model the data for a single weapon - possibly two
if there are ahead throwers (these bear forward, DCT and DC from racks "bear" aft) - and the number of shots is simply
the number of patterns - not the number of DC or tubes etc.

The way to have your cake and eat it too - get the ammo consumption of the entire pattern and the effect of only one round
able to hit the target per "shot" - is to use statistics. The chance a pattern hits a submarine is well worked out in theory -
and the difference between that and practice is well understood as well because we have good statistics. What we do not have
(I think) but might be able to derive by testing is the % of a "shot" hitting in game terms. The % is almost an overstatement -
in 200 attacks (mainly by DC and Squid) in the Falklands, 100% missed - by the NATO specialists in ASW! [To be fair - only
a couple of the attacks had a real target - probably a lot of whales were attacked as well] But if you want to say that a good value
is in the single digit % range - that is reasonable and more fun than a fraction of a % - then you just make the value of your
weapon be n - and a ship with a larger pattern gets 2n or 3n - etc.

If a Tacoma has 4 K guns on one side, and one wanted to model that with centerline Y guns, surly one would use 4 (each throwing 2) to the
one a K gun throws. What bothers me is that a submarine is not really attacked the way a surface target is - on the bearing of the weapon.
It literally moves into the weapon that hits it - whatever its bearing was when the weapon was dropped. The attacking vessel essentially
must close to zero range and then the attack occurs astern in the age before ahead throwers. Except for Squid, one ought not think of firing
in the direction of a target off the side of the ship. And in order to hit, Squid tries to surround the "datum point" with DC - so that whatever the
target does - it is likely to be close to one of them. It just doesn't have to pass over the datum point before it drops like a classical ASW ship did.
The point of the pattern is to get explosives in a circle (or oval really) such that the target is likely to be near one or between two - assuming
one had actually solved the fire control problem with good data (which is rare). But it is so statistical that it can be abstractly modeled fairly well.

[JWE]

ORIGINAL: Local Yokel
It seems to me that it also makes it clear that you have little to fear by restoring to those E's their full complement of 12 or 16 side throwers provided you break that complement down into multiple slots. Adopting JWE's example, if a C- or D-gata kaibokan's 12 throwers are broken down into four groups of 3 (2 on each side) as suggested in my post #13 then each slot (if it fires at all - JWE's 2a) has a 2^3 % chance of a hit = 8%.

Two things, John. Please don't make the mistake of thinking that's how the code works. As I tried to make real clear, it doesn't signify. It was a quick, simple, and easily understandable way of showing magnitudes. It might serve as a good broad brush rule of thumb, but it has no mathematical applicability to the algorithm.

The other is that the probabilities have depth as well as width. 4 groups of 3 are definitely not as powerful as 1 group of 12, but again, the total of the sequential hit probabilities of dropping 3, 3, 3, 3 are way more than twice (almost an order of magnitude) the the total of the sequential hit probabilities of only dropping 3, 3 (but at least it isn't 4000 times [;)]). Not to say you shouldn't try it your way, but you may find you have only gotten part way there in getting those pesky Es into line. Something to keep your eye on, anyway. Give it a shot, let us know how it works out.

[JWE]

Btw, Mr Palmer, we are vetting "every" combatant ship class and adjusting "everything" to the model: Sum[Num x Ammo] = (Irl Ammo)/2

While we are at it, we are getting the launcher distributions (# of slots, Num per slot) for everybody down pretty pat. Have good data on who did what to whom and when, so we'll tweak the Tacoma/Colony class while this is going on. The project does not seem to require any major adjustments, just a lotta little dinky tweaks here and there. Nothing especially noteworthy, except for ammo, but what's good for the IJN goose is certainly good for the Allied Gander.

While we are slicing and dicing the class slots, thought we might get the DD radars under better control, too.

Coming soon to a theater near you.

[Buck Beach]

ORIGINAL: JWE

Btw, Mr Palmer, we are vetting "every" combatant ship class and adjusting "everything" to the model: Sum[Num x Ammo] = (Irl Ammo)/2

While we are at it, we are getting the launcher distributions (# of slots, Num per slot) for everybody down pretty pat. Have good data on who did what to whom and when, so we'll tweak the Tacoma/Colony class while this is going on. The project does not seem to require any major adjustments, just a lotta little dinky tweaks here and there. Nothing especially noteworthy, except for ammo, but what's good for the IJN goose is certainly good for the Allied Gander.

While we are slicing and dicing the class slots, thought we might get the DD radars under better control, too.

Coming soon to a theater near you.

Probably not something appropriate for this thread but something to think about if your going to do some tinkering away. I have always thought there should be some way to account for those naval guns connected to a "fire control system". Maybe an accuracy penalty (separate devices) for the merchant ship's guns or other ships accordingly not having such systems.

Not smart enough to figure it out beyond the thought.

Buck

[Local Yokel]

John, you are right to chide me for over-simplifying. And I take your point about the rate of increase in sequential hit probabilities. Nevertheless, the larger the number of weapons an ASW ship can put in the water on a single pass, surely the more effective its attack is likely to be?

Elsewhere I've touched upon IJN doctrine being the laying down of the biggest possible pattern on a CertSub, and the USNTMJ's report’s description of a 19-charge pattern for a kaibokan. It is with this in mind that I should like to find a way of imparting to the IJN E-class ships the benefit of their large number of throwers, provided this can be reconciled with the need not confer Wunderwaffe capability upon them.

In the case of 6-per-side launcher banks we don't necessarily have to split them into 2 banks of 3, and presumably a reversal of that arrangement to 3 banks of 2 per side will further attenuate an excessive increase in sequential hit probability whilst preserving the full number of launchers. Doing this would require the use of 4 extra slots, but even in the worst case of the Ukurus there are 8 spare slots available (the Ukurus are the most difficult to model in the conventional way, given their fit of 16 Type 3 launchers).

If some further toning down of the late kaibokan classes is required, then what about the capabilities of the depth charge devices themselves? In this thread you explained that a depth charge device's Range value was a reflection of the maximum depth setting of its pistol. In stock scenario 1 Range values for Japanese d/c's correspond closely with the max. depth settings given in Campbell (T-95: 197ft (game:164); T-95 Mod 2: 295ft and T-2: 476ft), which seem to bear this out. However, in Big Babes these values have been changed to T-95: 175; T-95 Mod 2: 275; T-2: 357 and I must say that this is a bit of a puzzle, particularly since the 357 value for the Type-2 doesn't seem to correspond to any of the pistol depth settings given in Campbell. Does this mean that the depth charge device's Range value is actually referring to something other than just depth?

What occurs to me is that if Range does represent max. depth pistol setting, then we have a means for limiting the depth at which the can will explode, below which the submarine may successfully evade. For example we could have a 'Type 2 d/c Shallow' device having all the characteristics of the standard Type 2 but with a Range of, say, 98 for the 30m pistol setting rather than the standard Range of 357 (or should it be 476?) If you make a proportion of the multiple slots on the kaibokan fire this shallow setting charge, you have enhanced the likelihood that the attacked submarine will have evaded to a depth where it is safe from such a weapon's attack. This might be a way of simulating the shallow-set charges fired in a large pattern – see the pattern diagram in the USNTMJ report - and should further tone down the potency of the kaibokan if that is necessary.

I'll post separately on the subject of radars as this was something I took a look at in the kaibokan TROMS. Will just say here that I am delighted to hear that you are looking at launcher/ ammunition values across the board.