ASW Thoughts (Full Version)

All Forums >> [New Releases from Matrix Games] >> War in the Pacific: Admiral's Edition >> Scenario Design and Modding



Message

dwg -> ASW Thoughts (2/4/2014 10:09:16 AM)

I've been poking around the ASW parts of the devices database(s) and there seem to be a few holes, not sure if anyone's made an attempt at a systematic fix, but Babes and Stock seem to have some common issues and some distinct.

Effectiveness
It looks like effectiveness figures are based directly on the warhead weight in lbs, but they don't seem to have been corrected for the fact some are TNT, and some aren't. Torpex, for instance, has a relative effectiveness of 1.5 in comparison to TNT, which means some of the depth charges are being fairly badly undercut in comparison to their contemporaries, e.g. the US Late-War Mk 9 replaces the earlier Mk 6, but it's shown with an effectiveness of 200 versus the 300 of the Mk 6. That's a true reflection of their warhead weights, but (according to navweaps) the Mk 9 had switched to Torpex from the TNT of the earlier weapon, so they should actually have equivalent effectiveness (300x1 and 200x1.5). OTOH Squid is way out in Stock with an effectiveness of 35, which corresponds to the weight of a Hedgehog bomblet, the Squid bombs were actually 207lbs of Minol (anyone have a relative effectiveness figure for Minol?).

Accuracy
Some of the accuracy figures are distinctly iffy. Most DCs seem to have been given an accuracy in the 8 to 10 range, but the Mk 9 in stock has an accuracy of 30, it's quicker-sinking, spin-stabilised bomb should increase accuracy a point or two, but not triple it! In stock, Squid has an accuracy of 164, in Babes it's been cut back to 24 (though the effectiveness is fixed). Squid was reckoned to be around as effective as Hedgehog (168 in both Babes and stock), and Double Squid distinctly better (RN late war single attack kill percentages: DC 6%, Hedgehog 25-30%, Squid 25%, Double Squid 40%).

Range
Aka depth capability in feet. These mostly seem accurate, but some weapons had mid-war updates that increased their ability to target deeper-diving boats and not all of these are represented. One issue that caught my eye is that Babes has the Hedgehog cut down to 400'. It's a contact-fuzed bomblet, unless it's hitting it's crush depth I think it should stay the 1000' of stock (I'm perfectly willing to be argued out of this if someone can point me at a reliable reference).

What actually got me looking at all of this was trying to see if the UK Mk X (and it's 2000lbs of warhead) was in the database. Beyond the stats issues, the Mk X needed special racks for use on the Captains (it was initially designed to be fired from a 21" torpedo tube on DDs), and that got me thinking about racks in general, and the way the game models ASW weapons. There was a lot of scientific analysis on attack patterns, and an evolution of how attacks were conducted. Initial '6 pattern' attacks fired 4 DCTs and 2 racks during an attack, for 6 DCs, but this evolved into the 10 pattern, with 4 DCT shots, and 3 shots each from the pair of rails, there was even the 14 pattern, and the ex-Brazilian Hs, with 8 DCTs and 3 rails, could manage a 17 pattern. Eventually the RN seems to have settled on the 10 pattern as the optimum combination of single attack lethality vs depth charge loadout.

What this means for the game is that the ASW model probably isn't using the rails nearly as much as it should be. Rather than firing as often as the DCTs, they should be firing two or three times as fast. So there's an argument that rather than saying a ship has 2 DC Rails, we should say it has 3 attacks from each of 2 rails, it might look a little odd to see a ship with 6 DCRs in the database, but it should give a result closer to what we're looking for.

(N.B. this is all biased towards RN experience, I'm not sure whether the US attack strategies matched, and I suspect the IJN never really got far enough into proper Operational Research to realise they needed to alter attack patterns)



witpqs -> RE: ASW Thoughts (2/4/2014 1:22:37 PM)

One thing: regarding 'tripling of accuracy', don't be so sure that tripling the number results in tripling the accuracy in-game. The numbers in the scenario database are used by formulae in the game engine that the developers are contractually forbidden to reveal to us, except in certain cases (where they have clearance or whatever). It it very likely that the formula involved has some math that does not yield a straight out linear comparison like the one you are making. They needed to look at the result obtained by the initial value, then modify that value in accord with their understanding of how the relevant formulae works (since they do get to see it), then compare the result obtained by the new value with whatever result they wanted to achieve. And repeat as needed until they get a baseline value. In the case of this stabilized weapon, if it was a one-off that is the end of the process. If not, they had to do that for a variety of weapons until they got a formula they could use to generate values for all the models of that weapon consistently.

As far as number of weapons goes (the section where you talk about patterns and such), something similar applies, with the 'Super-E' issue being the big example. Because of the way the game engine uses the data, when the database had entries to 'look' like the weapons load of the Japanese E class, the results were beyond what the platform was actually capable of. So they had to change the number of rails/launchers to compensate for that.

If you're able to search it out there was a lot of discussion about these issues in the earlier Babes threads, and again maybe a year back and more.



Symon -> RE: ASW Thoughts (2/4/2014 5:12:38 PM)

WiTPqs is right on. Lots of people, over time, have come to certain conclusions about Gary Grigsby’s games, and they are out there. I will not reveal that for which I have signed a non-disclosure for, but there’s a lot of stuff, out there, if you know where to find it.

What you must remember is GG has been at this since the dawn of time. He has a toolbox of “things” collected over the decades and simply “chooses and drops”. He doesn’t use much multiple precision and doesn’t use complex math routines. He uses tried and true stuff from a simple ‘include’ call (to a basic library.h). His code is simple, linear, and fast; necessary for 10E7 calcs per turn. That’s why he is able to pump out games and make a profit from his work.

This is also why it is absolutely impossible to tweak WiTP AE in the ways many people wish. GG’s algorithmic concept and structure is fundamental to the game implementation. Doing it different means scrapping it and starting from scratch with a 6 figure budget. But, knowing how it is structured, makes it easier to get internally self-consistent. It is linear: add, subtract, multiply divide, and return a value; compare to the pseudo random seed generated for that turn and give a result. This is exactly why there was a “Super E” issue with Japan and why one doesn’t want to mess with numbers of launchers.

The (Naval) ASW algorithm is tripartite. One must acquire the target by evaluating DL. DL varies on a unit by unit basis and is informed (a bit) from prior DL. It is a decreasing chance. Once one acquires, one attacks. Once one attacks, what are the chances to hit? This is the poop in the pudding.

Attack profiles are by unit, on a device-by device basis. A ship acquires a sub and attacks according to the number of launchers it has on a one-to-one basis. It’s the hit algorithm, run as many times as there are platforms. So, if you know your math, you understand that’s a 1-(1-x)^k proposition. X is kinda, sorta what the seed does with the other numbers, and K is how many times it does it (number of launchers).

Pretty clear that the more launchers you got, the closer to 99.99% you get. With a simple seed of 0.20, you get a 99.97% hit rate for 16 launchers on a boat. Bad juju, but it’s linear and deterministic. That’s why I cut back the number of launchers on the ‘Super Es”. One must understand the imperatives of the algorithm and make adjustments accordingly.

I hope this helps.



dwg -> RE: ASW Thoughts (2/5/2014 12:43:32 AM)

I think you've both missed my point to a degree. I'm quite prepared to accept the limitations of the model we have to work with, but within those restrictions, the input data should still reflect reality - if all rack and thrower launched DCs have an accuracy figure in the range 8-10, but the Mk 9 is out at 30, better even than some ahead throwing weapons, then something is clearly off. It doesn't matter how the algorithm runs, but weapons that were historically more effective clearly need to have that reflected in the input data in some way, and weapons that were historically about as effective as each other clearly need to have figures that are around about the same. It's picking up the outliers that don't reflect that that I'm primarily addressing.

JWE's fuzzified description (pace NDAs) of the ASW algorithm essentially matches my understanding, but that doesn't mean we can't look at the way things are playing out in game and say 'I think we can get a closer match to historical results by doing X'. That doesn't mean anyone else has to do X, and it doesn't mean that X will necessarily work, but it certainly makes X worth trying. My reading around the subject makes me think we're undervaluing DCRs, because they actually fired as many as 3 times for every one time a DCT fired, and there's pretty much a straight linear relationship between weapons launched and the physical probability of getting one within damage distance, so I think it's worth trying to see if we can get better results by reflecting that in the device data. I take JWE's point that as we add attacks we approach unity, but that actually argues for reducing accuracy across the board, not for leaving things as they are. A DD with 4 DCTs and 2 DCRs executing a 10 pattern attack WAS more effective than one with 4 DCTs and 2 DCRs executing a 6 pattern, and we should have a way of reflecting that evolution in the science of ASW, but at the moment the DB doesn't.

If I was programming this myself I'd use a completely different approach based on a single attack calculated on the volume of water swept by the pattern and the likelihood of that intersecting the sub (which is where the ahead throwing weapons, particularly Squid, have their big advantage, their attacks aren't limited to getting their platform on top of the sub), but I'm stuck with the algorithm in game, and tweaking it as best I can. I'll make my changes and see how they work, if they do, good, if they don't, I'll change them back and let the problem sit at the back of my mind until another approach presents itself.



Alfred -> RE: ASW Thoughts (2/5/2014 1:23:39 AM)

Have you read this thread from 2011?

http://www.matrixgames.com/forums/tm.asp?m=2849954&mpage=1&key=asw&#2849954

As always, all the posts from JWE are must reads.  He does a good job in describing the inherent difficulties in trying to be "historically accurate" within the legacy game code.  Never underestimate the pressure of trying to cater to the vocal crowd who armed with secondhand knowledge usually gleamed from the internet or some inaccurate published author, just demand their version of reality be implemented, irrespective of the practicalities.

Of course that doesn't stop anyone from creating their own mod for their own personal play.

Alfred



wdolson -> RE: ASW Thoughts (2/5/2014 1:44:50 AM)

Basically if we had been writing this from scratch, we probably would have done many things differently, but we were starting with an already existing game engine and we were making what amounted to a large number of tweaks. Some of those turned out to be quite extensive like the new ship repair stuff, but it wasn't possible to rewrite everything.

I agree with you that the ASW model could have been done better. So could a lot of the other models. As many have pointed out, the land combat model, especially in Asia, is less than ideal. And there are other areas I would change too if I were rewriting the program. Ultimately it is what it is. It's better than WitP in most areas, but it isn't perfect.

Bill



DOCUP -> RE: ASW Thoughts (2/5/2014 2:44:51 AM)

The developers did a great job.  Still one of the greatest war games made.



Symon -> RE: ASW Thoughts (2/5/2014 1:42:13 PM)

Oh, golly, I certainly wasn’t suggesting that folks don’t try. Was just trying to explain why thinking in real-world terms doesn’t quite you where you might want to go. To help keep from reinventing the wheel.

All I can say is try it and see how it works. A lot of people would be interested in the results. As a caution, though, try thinking of “accuracy” and “effect” as merely text strings that identify a data field. Don’t think of them as defining accuracy or anything physical. That should free you from some unintended limitations.

Good luck, and I hope things work out as intended. Let us all know, however it turns out.

Ciao. JWE



EHansen -> RE: ASW Thoughts (2/5/2014 2:35:05 PM)

When I read discussions like this I feel that this statement by Symon:

As a caution, though, try thinking of “accuracy” and “effect” as merely text strings that identify a data field.
Don’t think of them as defining accuracy or anything physical. That should free you from some unintended limitations.

is one the most important things to understand. It would probably be an appropriate statement for many 'data fields'





Alfred -> RE: ASW Thoughts (2/5/2014 3:37:48 PM)


quote:

ORIGINAL: EHansen

When I read discussions like this I feel that this statement by Symon:

As a caution, though, try thinking of “accuracy” and “effect” as merely text strings that identify a data field.
Don’t think of them as defining accuracy or anything physical. That should free you from some unintended limitations.

is one the most important things to understand. It would probably be an appropriate statement for many 'data fields'





Alas, that is a constant refrain from Symon.

Constantly people read into text strings a meaning which is just not compatible with the code. No matter how many times he says it, there are repeat offenders who insist on ascribing incorrect meanings.

Alfred



Symon -> RE: ASW Thoughts (2/5/2014 4:55:12 PM)

quote:

ORIGINAL: Alfred
Alas, that is a constant refrain from Symon.

Constantly people read into text strings a meaning which is just not compatible with the code. No matter how many times he says it, there are repeat offenders who insist on ascribing incorrect meanings.

Alfred

Whoops got ahead of myself typing. Brother dwg is one of the smart ones. He's one that's worth talking with and letting him go forth.

Know folks like this are far and few between, but they do exist. Please don't stop bringing the idiots up short. I love your takes on things. Makes me feel like we got something right.

Ciao. John



Stvitus2002 -> RE: ASW Thoughts (2/5/2014 5:01:30 PM)

quote:

No matter how many times he says it, there are repeat offenders who insist on ascribing incorrect meanings.


In defense of those who question game results/actions, as players,all we have to
go by is what we can see. We cannot see what goes on "under the hood",or what the
various formulas are. I guess there are a few on these forums who can see these things,
but for the rest of us,all we have to look at is statistics. Beyond that,a "spray &
Pray" mentality is all i can do.


WO 0/0



Symon -> RE: ASW Thoughts (2/5/2014 5:16:55 PM)


quote:

ORIGINAL: Warrant officer 0/0
In defense of those who question game results/actions, as players,all we have to
go by is what we can see. We cannot see what goes on "under the hood",or what the
various formulas are. I guess there are a few on these forums who can see these things,
but for the rest of us,all we have to look at is statistics. Beyond that,a "spray &
Pray" mentality is all i can do.
WO 0/0

How wonderful for you. If all you have is statistics, then you have a perfectly deterministic game. If you merely wish to play the game by your standards, then "spray and pray" seems to be about right for people of your proclivities.

Pardon me for not paying more attention to your vaporings, I've got some Tuscan meatballs to make.



Dili -> RE: ASW Thoughts (2/5/2014 8:04:12 PM)

It seems the system works like air bombing concerning quantities dropped reaching too easily 99% hit rate, unfortunately.



wdolson -> RE: ASW Thoughts (2/5/2014 10:41:32 PM)

It's a probability calculation. If the accuracy of a weapon is 5%, that's .05. The probability of missing each shot is 1-.05 or .95. If you take two shots, both of which have the same probability of hitting, the probability of missing drops to 0.95 * 0.95 = .9025. You keep multiplying those .95s together and the probability of a miss gets pretty small pretty fast.

This is why planes carrying two torpedoes are so effective.

There are probably some better ways to do the math, but modeling reality can be very difficult.

If someone wants to take a shot at improving ASW devices, I encourage them to try. The entire community may benefit from a better mix of factors. Back in the WitP days the CHS team took on anti-aircraft and improved it somewhat. It led to further improvements in AA in AE.



Dili -> RE: ASW Thoughts (2/6/2014 3:37:09 AM)

It would improve with another calculation before at aircraft(or squadron for level bombing)/ship level. The question this will answer would be: Is the attacking ship above the submarine? is the aircraft aligned/timely dropped with the target.



Alfred -> RE: ASW Thoughts (2/6/2014 3:52:41 AM)


quote:

ORIGINAL: Warrant officer 0/0

quote:

No matter how many times he says it, there are repeat offenders who insist on ascribing incorrect meanings.


In defense of those who question game results/actions, as players,all we have to
go by is what we can see. We cannot see what goes on "under the hood",or what the
various formulas are. I guess there are a few on these forums who can see these things,
but for the rest of us,all we have to look at is statistics. Beyond that,a "spray &
Pray" mentality is all i can do.


WO 0/0


It is simply not correct to say us non devs only have "what we can see" to go by.

1. Yes, only the devs can see what goes on "under the hood" but there are a great number of occasions when the devs have provided "peeks" into what transpires "under the hood".

2. It is incumbent on anyone who believes they know better than the devs, to attempt to understand what were/are the design parameters and practical considerations which impacted upon the devs freedom of action. More often than not these are matters which are "above the hood" and therefore are fully disclosed, not being part of any NDA.

3. When people propose alterations, they really should fully understand first what exactly it is they are advocating. Too often the old Spanish saying "he heard bells ringing but he doesn't know from where" comes to my mind as half baked/understood proposals are put forward (invariably bypassing points 1 and 2 above). Often the proposal is presented as a solution to an "idiotic/illogical" dev implementation.

Now I don't intend it to appear at all as if I believe the OP falls, although many others do, into this category. I am aware that the OP has produced DaBabes quality standard mods and he does do his research. However it was possible that he had not come across the thread link I posted which clearly indicates the difficulties in altering ASW for general consumption by the AE community and thus the sort of compromises which are necessiated. Producing his own mod with the assistance of the insights provided by JWE/Symon is another matter altogether and one which might be successfully accomplished.

4. A close observation of what actually transpires in the game when combined with a proper analytical approach and understanding the above points, will result in informed positions. What is often passed off as observation and analysis of AE praxis is nothing of the short but ill informed wishful thinking. Not very scientific, not very precise, not very logical.

The common thread in all of this is that a non dev needs to devote a considerable amount of research to bridge most of the knowledge gap resulting from not having direct access to what transpires under the hood. But that gap can be considerably narrowed with the requisite effort.

Alfred



Alfred -> RE: ASW Thoughts (2/6/2014 3:56:03 AM)


quote:

ORIGINAL: Dili

It would improve with another calculation before at aircraft(or squadron for level bombing)/ship level. The question this will answer would be: Is the attacking ship above the submarine? is the aircraft aligned/timely dropped with the target.


That requires two different calculations as the air and naval ASW algorithms are different from each other. Furthermore they were the responsibility of two different AE development teams.

Alfred



dwg -> RE: ASW Thoughts (2/6/2014 4:42:54 AM)

quote:

It's a probability calculation. If the accuracy of a weapon is 5%, that's .05. The probability of missing each shot is 1-.05 or .95. If you take two shots, both of which have the same probability of hitting, the probability of missing drops to 0.95 * 0.95 = .9025. You keep multiplying those .95s together and the probability of a miss gets pretty small pretty fast. This is why planes carrying two torpedoes are so effective.


Yeah, hence my point about doing things as a single attack if I had my druthers. Multiple weapons tweak the probability of a hit upwards, but that base probability is a one time thing. It's also why there's a quantitative difference between depth charges and ahead throwing weapons. With a depth charge pattern you have to physically drive the ship over the sub (and lose the sonar track as you do so), with a fixed ahead-throwing weapon (Hedgehog and Mousetrap), you don't have to have the ship overhead and can attack a target you can actually see, though you do still have to point it at the target, and with the trainable ahead throwing weapons, mostly post-war: Squid, Mk 15 Hedgehog, but particularly the 360 deg training Limbo and Weapon Able, you don't even have to point the ship.



dwg -> RE: ASW Thoughts (2/6/2014 4:53:44 AM)


quote:

ORIGINAL: Symon
All I can say is try it and see how it works. A lot of people would be interested in the results. As a caution, though, try thinking of “accuracy” and “effect” as merely text strings that identify a data field. Don’t think of them as defining accuracy or anything physical. That should free you from some unintended limitations.


I may try setting up a simple test scenario and run it a few times just to see what the differences are in the results. As for not thinking of 'Effect' as anything physical, it's a bit difficult when the values are clearly warhead weight in pounds ;) OTOH 'Accuracy' clearly has to cover several different factors*.

* Which reminds me of my first job interview - I was asked to list all the factors you had to take into account a) bombing a runway and b) shooting down the other guy. We were there a while ;)



dwg -> RE: ASW Thoughts (2/6/2014 4:59:31 AM)


quote:

ORIGINAL: Alfred

Have you read this thread from 2011?

http://www.matrixgames.com/forums/tm.asp?m=2849954&mpage=1&key=asw�

As always, all the posts from JWE are must reads.  He does a good job in describing the inherent difficulties in trying to be "historically accurate" within the legacy game code. 


Not yet, but it's in the list of things to do! The ASW spun off from what I was actually trying to do and grewed and grewed like Topsy. The initial intent was simply check if the Mk X was there and add it if not, that turned up a few db anomalies, and the wider discussion of the ASW model is about 4 stages removed from what I was originally doing! OTOH its interesting enough I do think I'll carry through on the testing, and JWE's thread will be interesting for that. But for the moment there's other stuff to finish.



btd64 -> RE: ASW Thoughts (2/6/2014 4:31:46 PM)

This game, no matter what, is the best war game made, pound for pound, dollar for dollar, that I have played, Board or computer, in a long time. 35+ years. A loud applause from the crowd.[&o][&o][&o] The developers did good by me. With that, I am going to see if I can get my butt out of trouble with the babes iron nasty.[:@]
Cheers one and all [sm=00000436.gif]



Dili -> RE: ASW Thoughts (2/6/2014 6:49:33 PM)

quote:

That requires two different calculations as the air and naval ASW algorithms are different from each other. Furthermore they were the responsibility of two different AE development teams.


I am just giving ideas.



packerpete -> RE: ASW Thoughts (3/17/2014 3:35:34 PM)

Don't know if you have this info. but hope it helps. http://www.hnsa.org/doc/depthcharge6/index.htm



Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
0.96875