Bizarre Battle behavior Bug, possibly game breaking

[vaalen]

This is something I have noticed happen time after time in my British 1880 campaign.

If you have more than one force in the same area, the AI will attack just one of them, often with overwhelming numerical superiority, and decimate it. While the initial battle report shows all forces involved, the detailed battle report will show that only one of your forces was engaged, and when you examine the stacks during your turn that should have fought, only one will have taken any casualties.

The worst example came in my war against the Ottomans, who keep a huge overstacked stack near their capitol. I had two full armies besieging Adrianople, which together outnumbered the Ottomans. A third nearly full stack came in to reinforce the battle. The third stack was the only stack involved, and it lost all three of its corps, with only the artillery batteries surviving. The other two armies took no part in the battle or the retreat. This caused my National Morale to fall by 14 points, from 86-72, a disaster.

I combined the two remaining armies into one stack, with a 35 percent penalty, in desperation. The two stacks had no penalties before they were combined. The Ottomans attacked, but withdrew before combat when they saw one huge stack.

This cannot be right. I have fought many wars in the 1850 campaign, and all stacks in an area would participate in the combat. They might not arrive at the same time, but now multiple stacks in the same area never seem to fight in a battle. The whole point of marching to the guns fails when the relief force is the only one who fights, and gets creamed. I have had to make sure that no stacks can reinforce me if the Ottomans attack, which is absurd.

Any help or insight would be appreciated. This is almost gamebreaking, in a big war with big stacks.

[loki100]

unfortunately its WAD. There is, but I can't find it back, a very good discussion of the stack combat mechanisms in RoP - the same applies in PoN. It can happen in a multiple stack fight that the enemy targets one of yours and not the others. I had it as a text file but seem to have lost it (had to back up my laptop and a lot got deleted).

The one critical time it happened to me was also in a war with the Ottomans when they concentrated on 1 out 3 stacks and pretty much destroyed it. However, their army's cohesion was shot and the other two were retreating but still in the province, so I reset their stances to all out attack and they took full revenge.

Its a part of the targetting algorithm, the same that within engaged stacks determines which brigades hit which opposing brigades. I find it happens a fair bit in RoP.

[vaalen]

unfortunately its WAD. There is, but I can't find it back, a very good discussion of the stack combat mechanisms in RoP - the same applies in PoN. It can happen in a multiple stack fight that the enemy targets one of yours and not the others. I had it as a text file but seem to have lost it (had to back up my laptop and a lot got deleted).

The one critical time it happened to me was also in a war with the Ottomans when they concentrated on 1 out 3 stacks and pretty much destroyed it. However, their army's cohesion was shot and the other two were retreating but still in the province, so I reset their stances to all out attack and they took full revenge.

Its a part of the targetting algorithm, the same that within engaged stacks determines which brigades hit which opposing brigades. I find it happens a fair bit in RoP.

Thank you, Loki. I had not noticed it in the 1850 scenario, but I did not not fight many big wars in those campaigns. I also have not played much ROP. It would seem to me that a stack in the same province, with a good active leader, would take part, literally marching to the sound of the guns. But I will try the all out attack strategy the next time I have a chance.

It has happened to me before in the Turkish war. One time, I had a full stack, and a much smaller stack of siege equipment. The massive Turk army wiped out my siege equipment force, while my main army had tea, and then retreated without firing a shot.

I am sorry you lost your link to that discussion of stack combat mechanisms in ROP. If anyone knows where it is, a link would be most appreciated.

[loki100]

found it, its the text file in this post

but this is the key text:

== '''Commitment of stacks''' ==

Stacks are sorted into 3 types depending on their posture: Aggressive/Offensive, Defensive, Passive. A stack in an aggressive/offensive posture in the region of battle is picked in a purely random manner and then will seek a target. It will choose first a non-moving enemy stack which is also in assault/attack stance and which is not already significantly threatened (meaning the enemy stack is attacked by a number of friendly line elements which is less than 150% of the number of line elements it has). If there is none, then a non threatened, moving stack will be picked.

If there is none, then the procedure will be repeated by seeking a non threatened defensive enemy stack (first non-moving, then moving). If there is none, then the procedure will be repeated by seeking a non threatened passive enemy stack (first non-moving, then moving). if there is none (meaning all enemy stacks are 'threatened'), then one at random will be picked.

This concept covers the first stack picked. What about the next? If the stack is already engaged by an enemy, it will fight back against him, and that's it. If this stack is in defensive/passive posture, it will also activate all other friendly stacks (of all friendly factions present) in the region (i.e if the combat is still involving only stacks outside the structure, then units in the structure will take no part).

If the stack was not yet involved but potentially could be (i.e Defensive stacks waiting for their release), then it will first help a threatened stack, if there is none, it will pick a target, as was done for the first stack that entered the battle.

Army HQs will never attack by themselves and always fight in support of another stack.

Correlatives: It is possible that a very small stack may target a large one, or the reverse. This is working as designed, and represents the randomness of battles, where a small force will face the brunt of an attack, or where it is on purpose delaying a large force while others troops attack the enemy elsewhere on the battlefield.

key is that a stack in attack mode will pick a target, that target is then locked, if there any more stacks in attack mode they in turn will select targets - most times this will steadily draw in all the fresh stacks in a province (or marching to the sound of the guns) but sometimes one can fail, or a large stack pick on a small one etc, then you get the situation you have.

In the main, march to the sound of the guns works well in PoN (it should with telegraphic messaging etc) so my usual choice is to set up my stacks so they are more or less of a size. If its a big European war, I tend to use stacks of around 3000 pwr, this can cope with a couple of corps, some divisions, artillery and all the stuff you need to drag around. That way, if I get unlucky in the target mechanism, at least its not a small formation that is at risk of utter destruction that can be picked off.

[vaalen]

found it, its the text file in this post

but this is the key text:



key is that a stack in attack mode will pick a target, that target is then locked, if there any more stacks in attack mode they in turn will select targets - most times this will steadily draw in all the fresh stacks in a province (or marching to the sound of the guns) but sometimes one can fail, or a large stack pick on a small one etc, then you get the situation you have.

In the main, march to the sound of the guns works well in PoN (it should with telegraphic messaging etc) so my usual choice is to set up my stacks so they are more or less of a size. If its a big European war, I tend to use stacks of around 3000 pwr, this can cope with a couple of corps, some divisions, artillery and all the stuff you need to drag around. That way, if I get unlucky in the target mechanism, at least its not a small formation that is at risk of utter destruction that can be picked off.

Loki, thank you for finding that information and that link, it is invaluable!

But my situation did not work as described, unless I misunderstood something. The fourth paragraph states;"If the stack was not involved but potentially could be(i.e. Defensive stacks waiting for their release), then it will first help a threatened stack..."

In my example, my reinforcing stack was in aggressive mode, and the single huge enemy stack followed the rules in targeting it, but the two defensive stacks did not come to the aide of my reinforcing stack, which was certainly threatened as it was outnumbered 2-1 in line elements. Instead, the two defensive stacks just sat there and did nothing.

My reinforcing stack contained three full corps and supporting units, yet all three corps were wiped out in three battles which happened in the same turn, while the idle defensive stacks did nothing, rather than come to the aid of the threatened stack as they should have.

Do you have an explanation for this?

Unless I have made some mistake, this looks like a bug.

[loki100]

I wonder if in PoN there is a leader test roll? I've seen exactly what you are describing happen to my armies and it seems reasonable (columns did sometimes find something else to do than usefully engage in battle) but re-reading you are right that the simple target selection process is not enough to explain it all (it will explain why sometimes stacks that are of very different size will clash)

[vaalen]

I could see this happening once in a while. But it has been happening more often than not in my current war with the ottomans, which is why I think it is a bug.

Maybe there is a leader test roll, but this has happened to leaders with strategy rating of 4, who have a theater commander who raises their chance of activating by 1, or, in other words, 5 out of 6 chance of activation. Yet it happens regularly to these leaders.

[Pocus]

I'm surprised you get so reliably this ganging problem. The code has not changed on that since several games, and ROP or AJE players don't suffer from it, so perhaps there are a specific set of conditions that makes it more probable in PON, but this seems odd.

The rules are:

a) if a defensive stack is attacked, ALL your defensive stacks will help him
b) if no defensive stack is attacked, and there is a fight with a stack of your in offensive, then confirmed, none of the defensive stacks will help (they are 'in reserve').

Now, if the enemy has several stacks, he will target without preference your defensive or offensive stacks, so chances are that he will end up not only targeting your offensive stack, but also one of your defensive stack, thus triggering the others defensive stacks of your.

But what you say the enemy always focus exclusively on your lone offensive stack. Or perhaps you had a straw of bad luck and you got that twice, but don't make it a rule.

The thing is, yes, there is a risk, to have stacks with different postures treated as being '2 layers'.

Last point, if everybody is in defensive but a single stack of your in offensive, then confirmed too, your stack will fight alone all the defensive stacks of your enemy, and he will receive no help from his buddies...

[loki100]

thanks for that, it confirms what I thought was going on.

as to visibility. I sort of suspect the lethality of a PoN battle, if both sides attack and with large numbers, the losses are high, so if a stack is isolated it will take very bad losses whereas in RoP it might just lose a lot of cohesion?

[vaalen]

I'm surprised you get so reliably this ganging problem. The code has not changed on that since several games, and ROP or AJE players don't suffer from it, so perhaps there are a specific set of conditions that makes it more probable in PON, but this seems odd.

The rules are:

a) if a defensive stack is attacked, ALL your defensive stacks will help him
b) if no defensive stack is attacked, and there is a fight with a stack of your in offensive, then confirmed, none of the defensive stacks will help (they are 'in reserve').

Now, if the enemy has several stacks, he will target without preference your defensive or offensive stacks, so chances are that he will end up not only targeting your offensive stack, but also one of your defensive stack, thus triggering the others defensive stacks of your.

But what you say the enemy always focus exclusively on your lone offensive stack. Or perhaps you had a straw of bad luck and you got that twice, but don't make it a rule.

The thing is, yes, there is a risk, to have stacks with different postures treated as being '2 layers'.

Last point, if everybody is in defensive but a single stack of your in offensive, then confirmed too, your stack will fight alone all the defensive stacks of your enemy, and he will receive no help from his buddies...

Pocus, I have spent a lot of time playing Ageod games, since BOA was introduced. I have viewed hundreds, probably thousands of battles from this code. I own and have played every game you guys have published with this system. I have played PON more than any other game I have ever had, and have fought many large wars with many large battles. I have never seen this kind of thing before, except in my 1880 game of PON. It has happened not only twice, but at least seven times, maybe more.It is happening much more often than not. I will test this some more, But I tell you this is strange, and different.

And this situation is not limited to the battles where I have mixed offensive and defensive stacks. It has happened when all my stacks are defensive, as well.

I exclude battles where I sent two offensive stacks into an enemy province, and the enemy sent one stack. In those battles, you have explained how only one offensive stack will be attacked.

I am concerned about battles that I have had in this scenario where I have two defensive stacks, no offensive stacks, a single AI stack of doom attacks, and only one of the defensive stacks takes part in the battle, and gets crippled and almost wiped out, while the second defensive stack does nothing.

In fact, the AI seems to know it will attack only one of the stacks. In desperation, I combined two stacks, despite incurring a 35% penalty. There had been two stacks in the previous turn. The AI moved into the province to attack, then retreated without combat once it realized the stacks had been combined. When the two stacks were separate, they were fully commanded, with no penalty. The AI prefers to fight two separate, fully commanded stacks with no penalty, than the exact same troops, combined in one stack with a 35 % penalty. Does this make any sense to you?

I have never seen this before, not in my previous games of PON, not in my previous games of Rus, BOA2, American Civil War, Napoleons Campaigns, not any AGE game.

That is the only reason I am making a point of it. I have no idea why it has happened in this scenario, and nowhere else in my experience.

I do not like to be the bearer of bad news, but how else are we going to have a good game, unless we report things like this?

[Pocus]

I can assure you that, unless the code is self mutating, the AI is not looking at your combat penalties, stack by stack. She can assess a stack power, and combat penalty lowers its power, so perhaps this is what you mean.

[vaalen]

I can assure you that, unless the code is self mutating, the AI is not looking at your combat penalties, stack by stack. She can assess a stack power, and combat penalty lowers its power, so perhaps this is what you mean.

Well, that is what I mean. Since the AI can assess stack power, and penalties lower stack power, why would the AI prefer to attack two defending separate stacks that have a greater stack power when in fully commanded individual stacks, to attacking the same units in a single stack, which would have less power because of the penalties?

Unless the AI expected to fight only one of those stacks?

Which is exactly what it has done, several times, in the game at issue.

That is the question.

As for self mutating code, very unlikely, but I have read and seen a lot of science fiction where this actually happens. Does the word "Skynet" mean anything to you?

Seriously, I deeply appreciate your looking into this question, even when it makes no sense, a question that only the deranged mind of a veteran grognard could raise. And I apologize for any frustration or confusion I have caused, which was certainly not my intent.

[SonOfAGhost]

She can assess a stack power

Uh, oh, he's anthropomorphizing the AI.

Does the word "Skynet" mean anything to you?

Or perhaps AGEOD are working on a new title called Global Thermonuclear War.

[vaalen]

Uh, oh, he's anthropomorphizing the AI.



Or perhaps AGEOD are working on a new title called Global Thermonuclear War.

I think Pocus let something slip.Athena lives! And she is female.

[Jonathan Pollard]

Uh, oh, he's anthropomorphizing the AI.

Actually in the French language, everything is anthropomorphized. There is no word for "it" in that language.

[Pocus]

yes, that's why I use 'she' to refer to 'her', although I know that in proper English, 'she' should be a 'it'.

[Athena]

Don't change your habit, unless you don't value your PC.

[Pocus]

you are weird.

[Athena]

.. says the schizophrenic prehistoric squirrel?

[loki100]

.. says the schizophrenic prehistoric squirrel?

good to see you again ...

[vaalen]

you are weird.

Pocus, I would not mess with her. Athena, please note that I will always refer to you as "She", or "Her".

[Sir Garnet]

The key thing about targeting is that if the initial target stack is thrashed in one round then there is no further combat.

Whether a leader is inactive makes a difference as well.

I also learned the hard way about the advantages of moving in reinforcements in a face-off in defensive posture or have them seek to avoid battle.

Note that the regions are multiple days' march across. Marching to the Guns has a roll based on days' march, inter alia, with a 10 day limit. In a period where battles tended to last not more than a day or two, this is a very generous way of reflecting coordination of separate forces converging for anticipated battle. However, it remains safest to have your forces collected together before the fight, which the stacking mechanism represents. With an enormous mass of men, Supply and CmdPt penalties can kick in, but you still have an ENORMOUS mass of men to absorb losses and allow feeding in of fresh troops round after round and day after day in order to win an attrition fight.

It is a nice aspect of this combat system that a battle might go badly against the attacker for a few rounds but then turn against the defender who no longer has fresh troops to face further attacks.

It also allows a lot of clever and daring moves, even effective spoiling attacks (which are a total waste in many combat engines).

I found sometimes being clever isn't, but I keep trying.

[vaalen]

I now believe that the heavy headquarters bug, where units in a stack that has a heavy headquarters will fire only at range zero, had a big effect on what I experienced. Since my first stack did not fire except at range zero, they were wiped out so quickly that the other defensive stacks may have had no chance to reach them. In this game, most of corps had heavy headquarters, and I noticed that most of the combat took place at range zero.

I understand that this bug is fixed in the beta patch, and will be fixed in the final 1.04 patch. And Philthib has promised a quickfix.

I really think this explains the anomaly.

Pocus, thank you for your patience!

[Pocus]

ah yes, indeed this bug explains a lot of problems. The HQ will be fixed for next quick fix, due out very soon (probably today)

[vaalen]

ah yes, indeed this bug explains a lot of problems. The HQ will be fixed for next quick fix, due out very soon (probably today)

That is truly wonderful news!

[Kensai]

I'm surprised you get so reliably this ganging problem. The code has not changed on that since several games, and ROP or AJE players don't suffer from it, so perhaps there are a specific set of conditions that makes it more probable in PON, but this seems odd.

The rules are:

a) if a defensive stack is attacked, ALL your defensive stacks will help him
b) if no defensive stack is attacked, and there is a fight with a stack of your in offensive, then confirmed, none of the defensive stacks will help (they are 'in reserve').

Now, if the enemy has several stacks, he will target without preference your defensive or offensive stacks, so chances are that he will end up not only targeting your offensive stack, but also one of your defensive stack, thus triggering the others defensive stacks of your.

But what you say the enemy always focus exclusively on your lone offensive stack. Or perhaps you had a straw of bad luck and you got that twice, but don't make it a rule.

The thing is, yes, there is a risk, to have stacks with different postures treated as being '2 layers'.

Last point, if everybody is in defensive but a single stack of your in offensive, then confirmed too, your stack will fight alone all the defensive stacks of your enemy, and he will receive no help from his buddies...

This begs the question. Do the same rules apply for allied troops? I have never quite experimented if allies get MTTSOG when in a defensive agreement. Obviously, if they do, it is already brilliant! If not, hopefully a feature to be added in all future AGE engine titles.

[Pocus]

Yes, in relaxed MTTSG mode (i.e PON not ACW where you must belong to same army), any allied corps/army can help you, as if your own nation.

[Kensai]

Brilliant. I was hoping it worked for PON!

[vaalen]

Brilliant. I was hoping it worked for PON!

Kensai, I saw it work yesterday, when an allied corps marched to the sound of the guns and joined a battle one of my corps were in. Actually saved the day!

[Kensai]

I hope so. Otherwise my Greek army may soon been polverized by the Ottomans.