DC's House of Whacky Variable Testing

[Mithel]

Although I don't advertise it publicly I've been building a database of documention on my website so this knowledge isn't all lost and doesn't have to completely be redone by everyone. I've been documenting things for months. At times it's really annoying because just a few words from Paradox could save us days, weeks, months of testing and guessing.

- Mithel

 

[unmerged(14683)]

What we really need, is sort of page organized like old Havard's modding manual.

How about starting the intermod project on that matter? Not sure which site should it host, but I believe we could find some volunteers...

We all have lot's of very specific knowledge and experience with various aspects of modding, but when someone starts or joins the effort, Havard's page can be only used only on the very basic level.

 

[O'Donnel Aboo!]

Copper, thanks a million for saying that! As one of the fledgeling modder hopefuls who has read HAvard's page front to back, I am at a loss for the actuall application of a great deal of script. Where can we (the virgin modders) find info like this? A MODing FAQ?

I'm not a mathnik by any stretch, but i'd like to help figure out if values between 0 and 1 are effective. Just tell me what to look for!

 

[unmerged(19545)]

There you go, O'Donnel Aboo! A test you can perform, to the benefit of the entire world, or at least the modder community. In a game in progress, take note of some value, like Soft Attack, for several different unit types you have, like Mountain, Militia, and Infantry. Then completely close the game down, and go into the Units Folder, within the DB folder, and change the base SA for each of those units. Just to see how the game is reading them, I suggest adding .1 to one, .6 to another, and 0.01 to the third. Restart the game, and check those values again. If it is rounding up, all three will be up one, rounding down, none will be up one, rounding nearest, then the .6 will be up, but not the others, and if it is rounding up, but not checking past the first decimal place, then only the first two will be changed.

Heck, why not go for broke, and make one value a substantial negative, to see if it takes absolute value, or actually reads it. I don't know that the info would have practical value, but like I said in the first post on this thread, you never know what some brilliant modder -there are too many to name, though it wouldn't be hard to come up with half a dozen, right off the top of my head- will accomplish with something like that. There are plenty of clever workarounds out there, most of them unappreciated.

Tell us what you find, in as careful detail as you can manage, and it will be a definite improvement to our collective body of knowledge.

Soapy, I loved your discussion about air effectiveness in WWI. It wasn't new to me, you were preaching to the converted. I overstated the case, as I am wont to do, but I didn't break it. Aerial observation of artillery fire was done in the civil war, though by tethered balloon, not biplane.
Anyway, I was most gratified by your passionate defence of air power. I have read far too many posts on here stating that air power is over represented in this game. But lets not go there, at least not in this thread.

 

[Soapy Frog]

On the subject of fractional odds I posted some time back. The Game appears to round normally... 0.5+ up, 0.499- down. The game keeps track to the 3rd decimal place at all times.

The minimum combat value for any unit is 1. if you set a value to 0.499 or less the game will still count it as "1", even if the number is stored otherwise in the save game file. Unsure what will happen to negative values...

Only integers SEEM to matter for combat (though this is devilishly hard to confirm). Therefore the engine appears to round to the nearest integer and then use that integer for combat purposes (all the while tracking the decimal number to the 3rd decimal mplace as mentioned before).

This is a recap of information already found in this thread

 

[Soapy Frog]

Anyway, I was most gratified by your passionate defence of air power. I have read far too many posts on here stating that air power is over represented in this game. But lets not go there, at least not in this thread.

Ok... but just to say this one thing... air power IS somewhat over-represented in the game, but this is due to the extremely low AD values of units, that never really increase apace with improvement in tactical_attack.

So what would often happen is the superstack of 12 advanced tactical bombers could completely overwhelm the AD value of any 12 division or less concentration, and over the course of a couple of days, completely destroy it.

 

[unmerged(19545)]

Let’s resurrect an old thread on Air Power in HoI. Or start a new one. Air power doesn’t really do what it should yet, but more perspectives might help to get us there. For one thing, much of what I've read here doesn't even agree on what it should do, or how well.

Back to the topic.

So, a lot of the last few days’ posts have been very hopeful. Particularly the need for a better and more current website to host clear information on what the community has learned so far. Any concrete suggestions on that? I need to learn to websites for my job, but the time frame isn’t really going to be good. Does Mithel want to make it part of his site? Or someone else?

I originally imagined this thread as being a place we could all (by all I mean everyone who reads the thread. Guess I’m just naïve) put in a few lines of what we know, then I, or someone else if consensus is that my writing isn’t technically clear enough, could collate the lot into something suitable for the FAQ forum.

Not that everyone reads the FAQs, but enough do to make it worthwhile.

 

[unmerged(19205)]

i will be making a new website for my upcoming superscience mod soon(within the next 2 weeks). i can always piggy back the FAQ if someone can compile them for me. In the mist of fine tuning my tech, units and events so a little too busy to hunt down all the stuff for the FAQ.

am actually waiting for 1.06 before i finalise everything and the stuff i read here posted by you great guys keep changing my thinking of how to implement my stuff correctly. since even though it superscience, am trying to make it as realistic as possible.

also looking for a good free webhost as i am using geocities atm. so if you can recommend one that would be great.

 

[unmerged(14683)]

Let’s resurrect an old thread on Air Power in HoI. Or start a new one. Air power doesn’t really do what it should yet, but more perspectives might help to get us there. For one thing, much of what I've read here doesn't even agree on what it should do, or how well.(...)

Good idea, in few days we should have some good material to discuss...

 

[O'Donnel Aboo!]

Negative Integers

Gentlemen! I am very excited to post to you the results of this morning's testing while playing around with negative integers as combat values. It appears they DO make a difference! Check this out:

First of all, I used a contolled environtment, the same province, the same leaders, and the same troop make up for each battle. 2 Militia vs. 2 Militia. Using Italy in the 36 scenario, I attacked the Ethiopian province of Dagubur with the same two Militia units, under the command of Lt. Gen Bergonzoli, a Level 2 Offensive Doctrine general, to counteract The defender's level 2, Devensifve doctrine leader, General Desta Damtew. The only thing changed in each of the three battles is the SA value of militia units. Each battle lasted until the Italian troops' organization broke.

Test one: SA = 1
Battle lasted 42 days. At the end, Italian militia had Strangth of 58 & 58 (116 Total), and each had an Org of 5. Ethiopian militias had 74 & 65 for strength (139 Total), and their ORG never changed from the 30 it started with. Italian General leveled up on day 17, Ethiopian General on day 19.

Test two: SA = -1
Battle lasted 44 days. Italian final strength was 56 & 66 (122). ORG was 5.
Ethiopian strength was 70 & 72 (142). ORG never changed from 30.
Italian level up occured on day 17, Ethiopian level up on day 21.

Test three: SA = -5
Battle lasted 45 days. ITL str 54 & 59 (114), ORG was 5.
ETH str 75 & 73 (148), ORG stayed at 30.
Italian level up on day 17, Ethiopian level up on day 33!

Several things come to mind here. First and foremost is the slower accumulation of XP for defending generals. It appears that not only does time affect XP gain, but the quality of the troops attacking does as well! I had never seen that mentioned anywhere before.

An obvious difference is that the Italians did less and less TOTAL damage to their enemies the lower their Soft Attack value was. Also, Soft Attack apparently has no effet on defender's ORG (what does?).

The one number that does not seem to add up is the Italian total losses. Leaving the battle with total strengths of 116, 122, and 114, respectively. I can't figure this out at all. Maybe one of you mathniks can.

One final inside the numbers stat. In the last test, where the defender gained XP much slower, the lapse in skill really hurt them. They lost 15 points of total strength in 16 days before the level up, 21 points in 16 days while the Italian was more skilled, and 14 points in 13 days when the skills were equal again. That 5% difference in combat effectivity made a bunch of difference.

If this holds up under scrutiny, then it means technology can actually work backwards, in a way, which is something i am VERY interested in knowing about because of the MOD idea i have. I anxiously await your analysis.

 

[Soapy Frog]

An obvious difference is that the Italians did less and less TOTAL damage to their enemies the lower their Soft Attack value was. Also, Soft Attack apparently has no effet on defender's ORG (what does?).

This may be simply a fluke, since the amount of damage caused by a successful hit is variable (1-3%). So you could conceivably run two tests with the exact same stats and get wildly different results.

The one number that does not seem to add up is the Italian total losses. Leaving the battle with total strengths of 116, 122, and 114, respectively. I can't figure this out at all. Maybe one of you mathniks can.

IMHO the combat engine was using SA of 1 in all three tests, and your reduction of SA to -1 or -5 did nothing whatever (although the slower XP gain for the ethiopian leader is kind of odd). 61, 58, and 52 damage inlicted by the Italians and 84, 78, and 86 damage inflicted by the Ethiopians seems to indicate results falling within an expected range. For example, if the average damage the Italians are inflicting is 56 with SA 1 (for the sake of argument) then given the same number of hits you could expect then to inflict as little as 38 or as much as 75.

The thing *I* find inexplicable and weird, is why the ethiopian units did not lose ORG at all. That makes NO sense, since every hit the Italian units got through should have inflicted 1-3% org loss in addition to the aforementioned casualties.

Unless units can recoup org during combat, then its possible that the Italians were not doing enough damage to overcome the rate of org recovery of the Ethiopians... then that would make sense. Now.. can only the defender recoup org during battle or can both sides?

 

[O'Donnel Aboo!]

bummer

OK, Soapy, i'll test some more with the same values (I KNEW you scientists would want something silly like that!)

I did two other tests in the last hour; I will post those results now.

Test four: SA = .5
Battle lasted 45 days. ITL str 60 & 57 (117) and ORG 5
ETH str 72 & 71 (143) and org remained at 30.
ITL level up on day 17, Eth Level up on day 18.

Test five a: SA = 10
Battle lasted 13 hours. ITL str: 74 & 79 (153) ORG was 0 & 6
ETH str: 87 & 90 (177) and ORG of 20 & 25
THis battle started 1 hour before nightfall, so I started over with the troops arriving at dawn

Test five b: SA = 10
Battle lasted 13 hours. ITl str: 70 & 83 (153) ORG 0 & 8
ETH str 84 & 83 (167) ORG 16 & 14.

In boh SA = 10 Examples, the leaders gained XP at an equal rate. In Test Four, the Ethiopian leader was 6 XP points behind.

THis disproves the earlier thought that SA does nto affect ORG....get it high enough and it certainly does. It also shows that SA is an important variable for the DEFENDER, as they shredded the Italians in a matter of hours. I will go back and retest the first two negative values (-1 and -5) and I will also do multiple tests of SA = -100 to see if there is a great difference.

I see your point, Soapy, it COULD be treated as a "1". If it is, I'll be sad. I'll go find out!

 

[unmerged(14603)]

The thing *I* find inexplicable and weird, is why the ethiopian units did not lose ORG at all. That makes NO sense, since every hit the Italian units got through should have inflicted 1-3% org loss in addition to the aforementioned casualties.

Unless units can recoup org during combat, then its possible that the Italians were not doing enough damage to overcome the rate of org recovery of the Ethiopians... then that would make sense. Now.. can only the defender recoup org during battle or can both sides?

Although we were not able last summer to fully explain how ORG declines during combat, it did appear that ORG is not affected until hits per hour exceeds 3. In all the tests ORG did not decline when SA < 4 even when GD = 0.

 

[O'Donnel Aboo!]

Upon further reveiw

after my initial excitement, I now beleive that Soapy is right, and that anything less than 1, even negative integers, decimals, and a Zero value, will be counted as 1 by the engine.

However, I did discover some other things while testing:

Game Speed matters! I could not believe it myself, i had to go back and check at least 5 times. BUt you suffer far fewer casualties at Extremely Fast and Very Fast than you do on any other speed. YOu also gain experience slower at the two highest speed settings. Apparently, as the game skips through three hours at a time, it also skips all the stuff it is supposed to be doing! The following are mean scores for the same test as before, SA set to 1 (not that it matteres, apparently):

Extremely Fast: Total Unit Strength at end of battle: 170 (85 & 85)
Total Unit Strength for defenders: 183 (92 & 91)
Total XP earned: 89!

Very Fast: Unit Strength: 144 (73 & 71)
Defender Strength: 164 (83 & 81)
Total XP: 142

Fast, and below: Unit Strength: 120 (62 & 58)
Defender Strength: 148 (75 & 73)
Total XP: 190

THis knowledge is certainly helpful! This is done with Vanilla HoI, 1.05c

Also, you may notice I didn't mention the enemy XP gains. That's because i found what can only be considered a glitch. I was comparing my generals XP gains to my enemies all along. Because of the time it took the cursor to get from one leader pic to the next, i noticed this. My leader started gaining XP as soon as the battle began. However, the enemy leader did not begin gaining XP UNTIL I HELD MY CURSOR OVER THE PICTURE!

I'm not joking, I tested this several times, too. WHile my leader would gain his level on Day 17 like in every example (Except when the speed was set too fast) the other leader would fall behind. I thought this was a function of the negative integers (would have been cool if it was). But no, if i held the cursor over both leaders before unpausing and letting the battle get started, they would gain XP at the same rate (THey are both level two leaders, remember). In my last test, I didn't check the enemy leader until my org was already about to break. Sure enough, he was at ZERO XP, and started gaining as soon as i had put my cursor over him.

THis HAS to be a bug! No WONDER we kick the computer's ass all the time!

 

[Soapy Frog]

Although we were not able last summer to fully explain how ORG declines during combat, it did appear that ORG is not affected until hits per hour exceeds 3. In all the tests ORG did not decline when SA < 4 even when GD = 0.

Whoa, weird. Interesting. I wonder if that's intentional, or a bug?

 

[Soapy Frog]

Game Speed matters! I could not believe it myself, i had to go back and check at least 5 times. BUt you suffer far fewer casualties at Extremely Fast and Very Fast than you do on any other speed.

Crazy. Weird. Sheesh...

However, the enemy leader did not begin gaining XP UNTIL I HELD MY CURSOR OVER THE PICTURE!

WTF? Did you post in the Bug forum?

 

[unmerged(13914)]

I can partially confirm the game-speed thing. As I collect hour-by-hour data on battles, HoI will occasionally jump from 01:00 to 04:00 or make some similar leap. When that happens, it does appear as if steps are missed in the combat calculations, and thus casualties wind up lower.

Note that this is at lower speeds, below normal, and it is rare -- I've seen it maybe two dozen times, maybe not even that, in the past year. Not sure why it happens, but it does seem to correlate with there being large numbers of combats underway, or with my having incoming e-mail and other loads on the PC. It could easily happen more often as the game speeds up, and as it would probably vary a lot from PC to PC, systematic testing would be very tricky.

I can also confirm that org appears not to fall during some battles . . . but if this occurs, it's only on the winning side. I have never, ever seen org go up during a battle. It may be that org damage is nonlinear, that is, the ratio of org hits to physical hits isn't constant. Another case of never assume.

I can't confirm the leader XP thing. I have seen enemy leader XP rise without any prompting from me, dozens or hundreds of times, and I've never seen any change come from my hovering the mouse.

For what it's worth.

 

[Mithel]

DC, yes, I will continue to document findings on my website. I own my own domain (actually I own two domains) and have maintained it for nearly ten years now, so it's not going to suddenly disappear. If you feel my website doesn't have something documented that we've tested and confirmed please e-mail me full info and I'll update my site. Hopefully you'll all understand that I tend to be somewhat conservative to assure that the tests and conclusions were valid. I think we've all gotten excited before when we "discovered" something and then often found two days later that our test was flawed.

It makes perfect sense that if the code/PC can't keep up to the hourly processing that it "skips" hours (I've observed this myself) and it's very likely the way the code is written that by skipping those hours it is also skipping the combat execution (thus leading to longer battles (lower casualty rates)). It all depends on how they are writing their multi-threading code and it's long been obvious that their thread code is flawed and susceptible to unpredictable (crashes) results.

To really observe and test this we would probably need to load HoI up on an old 300 Mhz PC or so.

It's also very believable that the code awards a daily org gain and if losses are low enough that compensates for the losses and appears as an unchanging org value. It might be worth a bit more testing to see exactly how this is working but it may also be trivial enough (i.e. we can't do anything about it) that for those of us that have moved up to combat values in the 20s or higher that it's irrelevant.

- Mithel

 

[Soapy Frog]

Here's an interesting finding (maybe you guys knew this already, I didnt).

Ground/naval AA efficiency does not seem to be affected by weather/terrain/night.

Anybody know about this? Will test some more...

 

[Soapy Frog]

Other points worth noting:

Modifiers in the Save Game File

When the weather/terrain/night etc modifiers are initialized and placed int he save game file, the number that is shown is not calculated consistently for all variables.

For weather/night the number is the bonus/penalty which will be applied to the hard coded modifiers in the game (which are not shown).

Frex: if blizzard is -40 to attack, and the number for blizzard attack in the save game file is 10, then in the game units attacking in a blizzard will show a -30 drop in efficiency.

For terrain the number is the actually efficiency bonus/penalty which will be applied to the units in the game (when the variable is initialized, the hard coded modifier will appear in the save game file).

Frex: if mountains are hard coded -80 to attack, the number that will appear in the save game file on initialization will be -80 +/- any modifiers applied by tech received or set in the .inc file, and that number will be the actual applied efficiency modifier in the game.

Movement is the same. The hard-coded modifier will show up in the save game file.

Efficiency's Effect on Combat

On the subject of efficiency, I am unsure of how precisely it works to modify combat values in the end.

As an example, I have 24 German divisions attack 4 French MTN divisions in the mountains in a lvl 5 fort. All divisions had comparable combat stats. French efficiency hit 260%+, whereas German efficiency was around 45%, dropping to 35% when rain set in and 25% when a storm started.

However, despite an almost 6:1 efficiency advantage for the French which should have almost fully compensated for their numerical disadvantage, their Org broke first... German losses were high, however most divisions seemed to have not suffered much org damage, while some were at 0 org or close to it.

I am at a loss to figure out why this is, I will try to repeat the experiment and get some better numbers...