Slitherine

quote:

ORIGINAL: el cid again


The technical answer is that maneuverability seems to be a simple value.

quote:


A programmer suggested it was probably just speed. Turns out some times (in stock) it is - speed divided by 10. Other times it was apparently speed plus a fraction of ROC. But there is one other factor used: number of engines. 4 engine planes of all types, and 2 engine bombers and transports, REDUCED maneuverability values by division. [2 engine fighters and night fighters were an exception to this- I have no clue why?]. So technically a third factor is involved, but for a 1 engine plane you never see it because divide by 1 equals the same value.

quote:

For what it's worth,
when I did an extensive test on the Zero Bonus late last year, I found that the difference in combat with the full ZB compared to without the ZB - seemed mainly DEFENSIVE.

quote:

ORIGINAL: el cid again

QUOTE: However, I pointed out to you a while ago that your continued suggestions that your formulas produce some sizable number of matches with the stock data base manuever values for aircraft was not correct.

REPLY: Yes you did. You were wrong then and you are still wrong. I spent hours of time posting this - and it was recognized as such - with appropriate comments.

Why do you "hope" I will do so again? What was the point of saying that? If you cannot read the data for yourself - and cannot read it when posted - well that is not my problem.

quote:

ORIGINAL: el cid again

If you have something constructive to contribute - fine. If you want to know what the field means you also may ask Matrix. As someone posted, many of the values in stock were probably not calculated at all - but seat of the pants guesses in the context of certain calculated cases (the ones that fit perfectly). No great amount of time (because time = money) was spend on the data set. Neither in research nor in calculation. Ball park was good enough.

Later in time, probably others added things without even knowing the field definitions the originator had in mind. So they - like CHS folks - had to guess on the basis "it is slightly less than a Zero" or whatever. I have lots of respect for people working with severe time constraints, with limited data, and/or with no field definitions. I don't have time to argue though. If you don't offer a constructive comment - I will just ignore what you say. I am a big boy who admits mistakes, confusion, lots of other errors: I have no time for endless rehashing. The data matches the formulas the programmers say it does - near enough - and we need to act as if that is knowledge - if we want to understand how the system works. I can imagine no reason for them to lie either. You really can look this up in this forum - and it isn't going to change because you say it isn't so.

quote:

ORIGINAL: Bodhi

quote:

ORIGINAL: el cid again


The technical answer is that maneuverability seems to be a simple value.

Correct, from Mike Wood's database editor structure post:
unsigned char mvr[MAX_AIR];
So it seems to be a value between 0 and 255

quote:


A programmer suggested it was probably just speed. Turns out some times (in stock) it is - speed divided by 10. Other times it was apparently speed plus a fraction of ROC. But there is one other factor used: number of engines. 4 engine planes of all types, and 2 engine bombers and transports, REDUCED maneuverability values by division. [2 engine fighters and night fighters were an exception to this- I have no clue why?]. So technically a third factor is involved, but for a 1 engine plane you never see it because divide by 1 equals the same value.

If it were derived simply from other data already present, such as max spped or climb rate, what would be the purpose of storing it as a separate value when it could easily be calculated?

quote:

ORIGINAL: ChezDaJez

quote:

For what it's worth,
when I did an extensive test on the Zero Bonus late last year, I found that the difference in combat with the full ZB compared to without the ZB - seemed mainly DEFENSIVE.

I remember those tests and also felt that the bonus yielded a defensive benefit more than an offensive one. Funny as the Zero bonus is a transient increase in maneuverability.

Chez

quote:

ORIGINAL: el cid again
I think you have misunderstood what I said: I did not claim that my modified formulas fit the stock data set. I tried to explain how they work in stock - an entirely different thing.

quote:

In stock the A6M2 has a speed of 33x mph and a maneuverability of 33.
It was suggested by a Matrix programmer that maneuverability was probably related to speed - and in this case I got the idea it was 10% of speed. A number of other cases in stock also show 10% of speed. Other planes seem to add to this 10% figure a fraction of the rate of climb - usually ROC/500. RHS tried this function and settled on 5% of speed plus ROC/500 - in effect doubling the proportion of maneuverability related to ROC vice just speed. RHS does NOT use the mixed system of stock and CHS - where some planes get a maneuverability purely based on speed - other planes a composite value - the latter system is used in all cases.

Since RHS wanted to reduce air combat lethality, we wanted to reduce the maneuverabilty rating. Particularly after Joel Billings informed us that maneuverabilty was the PRIMARY variable used in tactical air combat code. So the fact a Zero has a much lower maneuverabilty rating in RHS was considered good. Since ALL planes are similarly reduced - and use a consistent definition - we feel we got a better relative result.

quote:

ORIGINAL: el cid again

I think you have a point. I do not understand it. Please amplify and clarify. What are "transient values?" And why does it matter if we cannot determine them?

quote:

ORIGINAL: el cid again

No point in going around again.

You are not required to believe what I was told - or what I see - or even each other (one sees 15% support for each formula, the other sees only one plane - while I see 1/3 within a point or so of both formulas).

quote:


You want to say I am "wrong" about this data because it is sloppy - but it isn't my data and I won't accept blame because that makes things relatively confusing: I did not invent this. For whatever reason you also are not interested that the model is working well.

quote:

ORIGINAL: el cid again

quote:

ORIGINAL: ChezDaJez

quote:

For what it's worth,
when I did an extensive test on the Zero Bonus late last year, I found that the difference in combat with the full ZB compared to without the ZB - seemed mainly DEFENSIVE.

I remember those tests and also felt that the bonus yielded a defensive benefit more than an offensive one. Funny as the Zero bonus is a transient increase in maneuverability.

Chez

IF the Zero bonus is as advertised,

AND IF Mike Wood is correct that the primary determinant of air combat success is maneuverability

then it is indeed funny - and it may be that the tests were affected by other factors.

The model is very complex - and also few testers use the same seed for each run - - so it is easy to get apparent results which are misleading.

It is likely that the effect is both offensive and defensive.

quote:

If Ki-44 was rated poorly in stock or CHS - they were wrong. Look it up. It is so clearly
better than German fighters they lost out after winning licence production deals: both
Me-109E and He-100 were licenced for Japanese production IRL - buit the decision to go with
Ki-44 is lauded by Francillon. The A6M2 is winning well in many tests - so it is hardly worthless -
and the Ki-44 cannot replace it as a long range fighter - ever. Anyway - I don't care if a plane
is better or worse - I am not a partisan. I do not "defend" a plane. It is what it is. Go argue
with Francillon. I won't play CHS or stock because they got the specs wrong on most of hte planes
-- and they are there for all to see and check. A professional programmer and mathmetician
verified my data - and it stands on its own merits - complete with citiations. One person objected
to my "changing the long range Pete" - but Pete was not long range - however passionately someone
may believe otherwise. I found horrible errors, sometimes overstatements of speed or ROC or
service ceiling by more than 100%. Ki-44 is great because of its ROC - its speed - and its
reasonable punch. It is weak in range - well you cannot have everything. Yes - I expect to see it as
a player preferred plane - because it should be - not because it is somehow wrong.

quote:

ORIGINAL: ChezDaJez

[quote Chez]

You state that the goal is to reduce air combat losses. There are many ways to do so such as increasing durability ratings. If you believe than the maneuver rating is responsible for excessive combat losses, why not use a multiplier and apply it to the stock ratings. Try a value of 0.5 or 0.75 multiplied by the stock rating.

REPLY: I do not believe maneuver rating is "responsible for excessive combat losses." I think grossly exaggerated weapon ranges, grossly (relative) firepower to machine gun ratings, and lack of ammunition limits, plus a routine issue when more than 50 planes appear on one side - are far more the issue. However, there is evidence that maneuverability ratings in the higher ranges may be a serious contributing factor. Your proposal would have merit IF stock (or CHS) values were reasonable in a relative sense - but they are not. It would, however, work statistically speaking (preserving the relative errors) - to the extent maneuverability is an issue. Mostly it is not an issue - because mostly ratings are not in the high 30s and 40s - and because other things are unrelated (planes can shoot 100 times if need be).

Let me blunt here. The formula that you propose to use for RHS is flat out wrong! It has more holes in it than the USS Arizona. Any formula that refuses to recognize the importance of roll and turn rates will never reflect reality when it comes to maneuverability.

REPLY: There is no data in the set for roll or turn rates. There is no way to get that data for all planes in the data set in any single source (e.g. Aircraft of WWII) or set (like Osprey books) which use the same standards of data. It is not part of the existing model - which works well as you admit - and no way to get the data to an honest and easily verified standard with a reasonable amount of effort. If there was a way - and please tell me I am wrong about no data and you have it all ready to post from IISS or any good place - we don't have a proposal for how to use it in the maneuverability field. This field - which may be misnamed to an extent - must always include what the model needs it to include - even if we modify it for your pet factors.

And I WANT to include your factors. Not as dominant - they are not. You are the one who is wrong when you say they are "the most important." The most important factors are not plane performance in the first place. When the guy we are attacking does not see us, and flies on fat, dumb and happy, your factors matter not a whit: and that is the vast majority case. When we see something we dare not engage (we being in a C-47 and they in Ki-84s say) - and we go hide in clouds and open the range so they don't ever see us - your factors matter not a whit. IF we use these factors, we need DATA on the portion of the time they DO matter a whit. Much more often speed is going to matter (as we dive with our P-38s on the vastly more maneuverable - in your beloved horizontal - Zeros) - and win or lose we are going to keep on going -
not try to turn and fight him if he is still there to fight. How often does horizontal maneuverability matter? You tell me - and it decides the weight it gets. This is how to do proper analysis. Don't be emotional about a datum.

If all that mattered was horizontal - Ki-27 will win - and we will lose. You are going to give Japanese planes an undo advantage if you make this the main thing. A Ki-43 will be a giant!

My goal here is not to offend or to piss you off. That does no good. My goal here is to offer constructive criticism, heavy handed though it might be. I will refrain from any more posts on this discussion unless invited.

REPLY: You are civil here. So you are invited. My frustration is that your criticism is too short sighted: you never go all the way to a solution. The whole concept of using your factos is moot UNLESS you give us the data - data to CHS/RHS standards - data ANYONE can find in a proper source in a public library. It is also useless UNLESS you can balance it - although I can do that for you if you trust me - with other things. Just because you like maneuveraiblity does not make it king - it isn't - and we learned how to beat more maneuverable planes in fact in this campaign. Before we did, the turning in maneuver worked for Japan even against better planes. There is a lot going on - and no one fact is king all the time. Help us compromise the factors - which cannot happen by making anything king. And we will use it. We will use it the miniute it is clear you got it right. No delay. No pride in our past work. Better data is ALWAYS in in RHS - right now.

Chez

quote:

ORIGINAL: el cid again

Drongo (what does that mean ? just curious)it isn't helpful to bicker

I won't

I also have made promises not to say every last thing - and I am sorry if being ethical
prevents me from clarifying every last thing as much as I might like

I love to get technical and to explain - it is what I do. I have gone as far as I can here -
possibly one or two steps farther than I should - and the journey is over.

I do think your interest is laudable. I do not think your opinion is unreasonable either.
But I am just the point - the guy out in front - taking the heat. It is not my data. It is not
my code. And I also do not know all there is to know about it. In the nature of these things,
maybe no one does know everything about it - it may well have evolved - and to some degree
I know this is the case here. If the poor fit of data with algorithms bothers you - well we are
much more in harmony than in opposition about such matters. If I had done the original data I
assure you it would be at least 90% consistent. I didn't. It isn't. It is confusing. But you cannot
understand where to go unless you know where we began and where we are. I tried to help you
understand - as I have been led to understand. I made nothing up. It is not my data, my theory,
my code. All I did was go on and make a modified algorithm. You got a better one - I will
instantly adopt it - and say it is yours for the rest of my life. You got better data - I will instantly adopt
it and say it is yours - until still better data comes along. That is as constructive as I know how to be.

quote:

ORIGINAL: el cid again

Chez:

What really matters in air combat is actual speed. What is actual speed? Whatever it is, we cannot put it in a field! It is anything but constant. The air model does this for us. How does it know what to use? It uses at least three fields:

max speed
cruise speed
ROC

It may also use altitude.

It also used die rolls. Each pass each plane gets a new positional record.

As outsiders to the code routine executing an attack, ALL we can play with is the input data.

It is actually silly to object as you do to max speed. Right or wrong, it is what is used. And code is right.

Think of it like this: Whatever the ratio of acceleration to max speed may be, at any altitude, it is related to the max speed
of the plane first of all. And the routine DOES know the altitude too. It ALSO knows "this plane was cruising at the start of the engagement" - or "this plane was on an attack run at max speed" - so it knows initial speed. From there on the routine tracks the situation - not us.

Not sure what you want to do? But complaining about how the routine works is not productive. Ain't gonna help nothin'
And it can be made to work. Run 100 tests with RHS. Surely it works better than CHS or stock - and slightly better
than Nick Mod. Not bad for a crude first guess. Now - if you can make it better still = tell me how?

I do not use sustained ROC. I use initial ROC - FYI. The routine is set up for that - using anything less is not going to model well. It reduces ROC as altitude increases - as it should.

ROC matters in the same sense speed does - it is a single value the code can look at - combine with other things - and play with. It turns out to be directly related to power loading and speed - as either goes up so does it. These are not bad things to be considering in an air combat routine.

I do not like, did not design in, and would not design in, no fields for various other factors. I myself have THREE different "agility" factors for every plane (at 5000, 15000 and 25000 feet - and sometimes at 35000 as well), plus wing loading, power loading, weight, power, ROC, a correction factgor for ROC, a dive speed, and a dive speed limit, among others, in my own models. If data was available - it isn't - I would have things like sustained turn rates - as a direct rather than combined or derived value. But I do figure it out - and it is in my agility ratings. What shocks me is that this - taking over an hour per plane to work out - only gives virtually IDENTICAL values in air combat to my planes. WITP works - better than I think it should work. Maybe we will always do it this way - since we can do a plane in about 2 minutes - and it works.

quote:

I do not use sustained ROC. I use initial ROC - FYI. The routine is set up for that - using anything less is not going to model well. It reduces ROC as altitude increases - as it should.

quote:

ORIGINAL: Herrbear

quote:

I do not use sustained ROC. I use initial ROC - FYI. The routine is set up for that - using anything less is not going to model well. It reduces ROC as altitude increases - as it should.

Thanks for clarifying where you obtain the ROC. What do you use for max speed? Many sources list various ways, some at sea level, some at other altitudes, some you never know when it was calculated.