Proud Words On A Dusty Shelf

[John Roddy]

The horse is already out of the barn on this and long gone, however, I keep encountering players in sidebar conversations who seem to be seeking confirmation that what they think they are perceiving is real or an explanation for why they are imagining otherwise.  I'll confine my observations to the Napoleonic series as it is the one I'm most familiar with.

I'll begin by simply stating that the thing that has caused me to be a loyal proponent of John Tiller's games since their inception has been that the games played better than anyone else's and that whatever that magic is in the way they simulated military history remained consistent throughout the passing decades.  That is, though there were many many bug fixes and subtle tweaking of the game engine, the basic way the games played remained the same.

The reality is that this is no longer true and many who are not as comfortable delving into the mathematical algorithms involved are telling me they nevertheless feel that underneath all the nice new graphics, user interface enhancements, and optional rules that they enjoy, somehow the games are not playing in the familiar way they've come to expect at some nebulous core level.

It has taken some time to test this experimentally but I must reluctantly confirm that  casualties are no longer determined in a strictly historical and linear manner for any given combat.  By that I mean, in the real world, if you add more fire or more men in a melee, you will tend to get a linear increase in the effectiveness of both overall.  The game engine has always produced linear results in the manner I am describing in the past, however, those days are gone.

Simply put, one now sees many examples in a test environment where the addition of additional units firing or meleeing instead results in fewer enemy casualties.  Somehow the algorithm has jumped the statistical track in a different way than just averaging a bell curve result.  Instead, various factors in the fire and melee calculations that have no real world counterpart are now the most important thing in determining the result.

I have experimentally found a number of them but will not list them as I am certain there are more.  The net effect is what matters, it is far too often the case that withholding fire by certain units from firing or meleeing for a specific combat instance produces a more favorable result.

Once I verified this, I ceased testing as to whether it might eventually even out by the end of the game as the unhistorical absurdity of what is going on has already confronted me with a gamey abstraction that didn't used to be there.

So, to those who keep wondering, yes, the 4.0+ games are now different enough that quantifying exactly how different different really is just a replay of the question of exactly when the Ship of Theseus does or doesn't become a new ship.

[Dion]

That's why I don't want them to start selling games on Steam, it'll probably just make things worse. Though, I haven't played any 4.0+ games yet, so I'll have to hold off from being too critical, but I see what you mean.

More disturbing, is this new world wide AI. What's going to happen when that thing goes on-line? Will it act like a computer virus? It might end-up being like the old DOS days all over again, when DOS became obsolete, but this time it will be Windows instead. I hope not. I don't know about you, but I don't want to have to go through all that again.

[Warhorse]

What game(s)?

[Carlos]

Simply put, one now sees many examples in a test environment where the addition of additional units firing or meleeing instead results in fewer enemy casualties.  Somehow the algorithm has jumped the statistical track in a different way than just averaging a bell curve result.  Instead, various factors in the fire and melee calculations that have no real world counterpart are now the most important thing in determining the result.

John, as the saying goes -- extraordinary claims require extraordinary evidence. Pardon the hyperbole, but if you have indeed identified issues like the one quoted above via empirical methods, I urge you to write to HelpDesk@wargameds.com and provide a precise description of the observed abnormal behavior(s), accompanied with supporting numbers and files generated from the current publicly available versions of the unnamed affected game(s).

Let me stress that I am in no way dismissing your claims outright, only pointing out that making sweeping statements on the General Discussions public subforum is not the best way to report concrete issues affecting a core gameplay function. As I myself have done in the past, I would suggest that you use the proper channels to present your case with enough backing data that a proper investigation can be undertaken and a conclusion reached as to whether we're dealing with a broad dissatisfaction with a particular design decision, or with an actual anomaly in the current implementation of an intended behavior.

[oz77]

... casualties are no longer determined in a strictly historical and linear manner for any given combat.  By that I mean, in the real world, if you add more fire or more men in a melee, you will tend to get a linear increase in the effectiveness of both overall.  The game engine has always produced linear results in the manner I am describing in the past, however, those days are gone.

Simply put, one now sees many examples in a test environment where the addition of additional units firing or meleeing instead results in fewer enemy casualties.  Somehow the algorithm has jumped the statistical track in a different way than just averaging a bell curve result.  Instead, various factors in the fire and melee calculations that have no real world counterpart are now the most important thing in determining the result.

In N-series musket fire, the number of men in the firing unit does not have a linear correlation with the FP and accordingly the losses incurred. There is a valid explanation for this based on the number of companies per firing battalion and their size relative to the 100m hex frontage. This was a smart and well justified addition WDS did in recent patches.

The same goes for artillery for stacking above 8 cannons per hex.


Regarding melee combat, if I understand you correctly, you're suggesting that the distribution of losses is no longer strongly correlated with the number of participants in the melee. If that's the case, analyzing such correlation and drawing conclusions should be based on methodical data gathering to ensure a sufficient statistical sample, followed by testing and analysis.

Since I haven't done any of that in a methodical manner, I cannot confirm or contradict that the correlation of casualties-men in melee is gone now.

But what I occasionally do is I run a small excel based calculator based on melee losses formula outlined in the manual to determine melee odds. Overall my observations are in line with the melee formula, even though I sometimes observe unexpected results in melee combat (similar to real-life engagements) but all of it still being within the Low and High combat values mentioned in the manual.

If there is mathematical evidence indicating a broken correlation between the number of participants in melee combat and the losses incurred, then perhaps there might be a slim possibility of a bug which could be reported.

For win-loss assessment, As far as I know, the exact formula behind melee win-loss outcome is not public knowledge and was not specified in the manual intentionally. However, based on my observations, the general principle of a 3-to-1 advantage for successful melee outcome seems to hold true.


___________

Extending the topic further, I believe, in fact, there should not be a linear correlation between casualties and the number of men in the hex in the first place

To my knowledge, and others may correct me, melee combat is managed through two formulas running in the background:

1. Casualties formula - as outlined in the manual.
2. Win-Loss assessment - an unknown formula.


Typically, in melee, only the front two or occasionally three rows of men would directly engage the enemy. Using an example of 1800 men per hex attacking in columns and considering the 100m hex space and historical formations, it can be estimated that roughly only ~400 occupying the front of the hex would be directly engaged. The majority of the 1800 infantry stack (~1200) would be formed behind the first three rows and would be masked, unable to contribute to enemy casualties.

The same principle applies to cavalry, at max only about 300 horsemen could fit into 100m front space and being historically deployed into two rows (with small and rarely used third row). Consequently, theoretically only around 300 could engage in combat upon impact. In reality, even fewer would actively engage, and typically only the front row systematically had a chance to use cold weapons. However, potentially more could engage during the pursuit phase, depending on space, terrain, etc. Adding further 300 charging troopers in the same hex making it to 600 would not be increasing enemy casualties. Such formation would be sort of regimental column which was more of a battlefield maneuver formation but was rarely used for charges due to its ineffective use of the squadrons at the back.

Accordingly, my estimate that the linier correlation between men involved-to-casualties caused should stop at:
1. ~400 for infantry
2. ~300 for cavalry
And after these thresholds the casualties should not be linear.


There is another consideration of Win-Loss assessment which is currently handled by a separate formula (I believe). This formula should certainly be influenced (and most likely it is already) by the number of men in the attacking stack as seeing a large enemy force charging your position should be intimidating enough to trigger flee-or-fight assessment. In Napoleonic warfare this was mostly flee decision and rarely fight, as hand-to-hand close combat was a very rare sight.

Also, I would throw in another factor into the win-loss formula to represent situational awareness using threat values. As this value should be one of the key drivers for win/loss consideration in my view. And the threat value may come from all other enemy hexes, not just from the hex of the attacking stack. Seeing large enemy force approaching you should be impacting the morale and motivation to hold position despite the odds. Even though rationally thinking it is clear that enemy soldiers of the third row and further 200 meters to the back, will not be using their weapons on you it is still morale breaking finding yourself heavily outnumbered.

If the threat value would be considered in melee formula, it then makes sense to concentrate large cavalry, formations like commanders did historically and to maximise threat value. As of now there is no value in replicating massive cavalry charges of thousands of cavalrymen (think of Eylau!). An entire Cavalry corps of any size can be stopped for good by a couple of 100men squares. The tiny squares have no fear or concern of the massive cavalry body in front of them. Only the attacking hex is being considered with numbers within the melee they fight, regardless of the wider context. No situational awareness.

So summarising my opinion of the melee calculation - it does work correctly in my view and there is a room for further improvement via:

1. casualties formula can be adjusted via capping/non-linear correlation post certain threshold representing of theoretical number of troops able to use their weapons over 100m front.
2. win/loss formula could be adjusted with threat value from both sides taken into account, with the threat value coming from all hexes, to replicate the situational awareness.

[John Roddy]

To address the issues posed, (Warhorse) I am referring to all of the games in the Napoleonic series but I strongly suspect that some of the same issues may apply to the American Civil War series too, and if that is true, it likely affects other series as well. That is why this is posted in the General Discussions section and not in the Napoleonic series section.

(Carlos, Oz77) We cannot have the discussion that your questions would inevitably lead to or it would strain the goodwill of the forum operator (WDS) as to discuss this in sufficient detail to support my observations further would require me to mention things publicly that would not be acceptable here. My apologies for having to refrain from going into this in greater detail, it is not from lack of ability to do so, but rather discretion, as otherwise, this thread would go missing in action or further commend ended.

(Oz77) Please tread lightly on the Help Desk thing, there is no need to waste the Help Desk's time. I have to dance around exactly what I am talking about but I assure you that WDS is aware of their game engine design decisions that have produced what I have observed. The only possible exception to that may be whether they are fully aware of the extent statistically that certain changes have produced, i.e., nary so much a bell curve or truncated (averaged) bell curve as, well, something else.

(Carlos, Oz77) I can see that I have not quite communicated successfully what I am observing and so all I can really say is that we have sort of an apples and oranges situation and though your points are well taken, they really pertain more to something other than what I have observed. My apologies for that.

Please note, the purpose of this post is not to advocate a change or to prove my assertion or to defend my analytical abilities. Rather, I am simply here to confirm to those that have already glimpsed the elephant hiding in plain view that they are indeed seeing an elephant. I am not here to prove to others that there is indeed a hidden elephant lurking about underneath the hood every time they conduct fire or melee combat. If you don't quite see what I'm talking about, well, I think that is probably for the best ultimately.

If you wish you may contact me privately if you've just got to know, but, be aware that should I manage to explain it better, my prediction is that you will agree with me that you would have be better to have just written me off as a bozo and dismissed my observations as having no merit. Enjoy the new 4.0+ games, don't risk your enjoyment looking under the hood unless you too, like I and the others this post was written for, just can't help noticing things have changed, and not for the better.

[Dion]

Napoleonics, that explains everything. Abnormal casualties are caused by the extra time it takes to load muskets. A musket is only a fraction as powerful as a rifle.

[rahamy]

John,

Carlos is with WDS Support. You may do as requested - substantiate your claims and submit them formally (using current versions of the games) to Helpdesk@wargameds.com - and we will review them, and if there are in fact problems we will resolve them.

Your comments here are not acceptable however, and come across as the company is doing something nefarious - which it is not. Accordingly this topic is now closed.