Another impossible-to-loose battle lost (5 battle bugs)

[lostangelonline]

This is actually the very next turn in my 6-months MP game, after I lost the battle described here. So what happened:

1. I have 2 armies in my Plain-terrain region; I give them commands to each attack+siege 2 distant enemy regions (to NE and E);

2. Before they managed to move out from the Plain, they are attacked and defeated by enemy army (the attacked detailed in previous post); but they still "retreated" to destination and conquer target enemy regions;

3. In the next turn, I give commands to both those armies to go back and "revenge" that battle I should have won (yes, my armies took damage, but also did the enemy), but this time I split biggest army and attack with only the best 28 troops (4HCav+10elephants+14archers, the rest will go to a region nearby to be used in attacks after). I commanded them to march at the enemy army, and also to Ptolemaic(also enemy to me and Seleucids) region behind it, where I suspected enemy will siege (the screenshot is before I gave commands, so I drew the lines):


4. Processing the turn, enemy indeed sieges the Ptolemaic region behind, while my troops are advancing towards it:


5. My best army arrives first to the Plain where I lost previous battle, and automatically conquers the region as Attacker, as it is supposed to:



6. My best army arrives to the Ptolemaic region where it engages in battle with the Seleucid army who was already there, sieging the Ptolemaic fortress. BUT SURPRISE (first bug): I am the defender!!


7. ANOTHER SURPRISE (second bug): 2 of my chariots are not in green margins, but replace 2 elephants (who replace 2 archers), leaving my flanks exposed and with weaker support. Therefore instead of winning, as Defender and messed up deployment, I only get a draw:


8. My second army joins the battle, and I finally get Attacker status. I will definitely win now.
EDIT: (5th bug I missed) my better 1-attacking-star general is replaced with weaker 0-attacking-star general.


9. AND LAST SURPRISES: the very weak cav replace almost ALL the much more powerful (as attack damage, even if low in health) elephants and HCav (third bug), which in turn replace archers as (weaker) support units(4th bug), lowering my total attack damage (ranged+melee) to almost nothing! Therefore I loose the battle which was impossible-to-loose, if not for these 4 bugs:


Please help me understand if this is realistic, or should be fixed. Thanks.
(FYI: It took me 2 days to print-screen and report these issues in these 2 threads (with edits, replies to comments), I just hope is worth it)

[Gray Fox]

In the first battle result, Dhavalaca's army wins, but two Elephants are damaged. You were the defender because, as you posted, you arrived before the Seleucids.

In the second battle, because the elephants are damaged, they end up in the second line.

The chariots have filled in for the Elephants, because they have a better attack than the archers.

Now, you don't have enough flanker units. The flanks lose.

The damaged elephants have a ranged attack, and fill into the second support line, removing two archers.

The battle ends in a draw.

In the last battle, somehow Anubhavya is in charge, with attack 0. So no attack bonus from the leader. The Seleucids get +1.

I can only see that most of the Elephants and the chariots are in the second line. I suppose they were damaged as above. That's the problem with a unit that only has 2 HPs. They don't have the stamina to stay in the line.

Light cavalry take the place of the Elephants and chariots. It looks like five of them lost in combat or pursuit.

I can only tell you that Heavy Infantry are better than Elephants. Heavy Cavalry are better than chariots. Here's the army I used for Rhodes and I won 90% of the battles.

[storeylf]

Obviously I can only speculate, and I dare say so can the devs without a save file, but:

Regarding the surprise defender, I think you have hit one of those issues where it is not clear cut as to what you get. I was discussing this on another thread (which I now realise was a FOG2 forum as it involved both games). Who gets to be defender is less than predictable at times and hence makes the 2-0/0-2 leaders very dodgy as you can unexpectedly be on your 0 rating. It is worth bearing in mind that in that other discussion the OP wanted to always be the defender in his territory - which would have applied here (actually no it wouldn't, I misread who owned the region). For myself I've modded the game to represent the 'best' leader (accounting for various advantages) getting to choose what looks best for him, and only dropping to the default logic if there is a tie in all those factors.

I think what may have happened to you is that after working out that no one is actually attacking (sortie or assault) and both have just arrived (so no clear defender) that you hit the random default case. Where a battle has no well defined defenders involved then side 1 (who ever that is, I assume pretty random) is the defender.

Regarding the chariots and elephants, this I assume is a factor of you trying to be maybe a little too clever on force and get the perfect army for 28 squares, but not accounting for battles en-route or deployment rules (ssome key stuff in your case is reasonably well explained in the manual).

If you look at the battle you had before this one, you will see that you have 2 elephants that are injured. These 2 elephants are the ones that go into reserve/support to start with, they won't go straight to the front line. That means that the 4 chariots are not all spare to go to the flanks, the flanks are only actually filled if the frontline is filled by other stuff first. So you fill the frontline with the elephants and chariots and then have 2 chariots spare for the flanks. The injured elephants are in support displacing the archers.

In the final battle, well you seem to have understood that one - the elephants simply don't like going on the frontline if they have even a single point of damage. That throws everything into disarray. Which Elephants were known for

NB not saying that is good or what, but that is how it works. Had you not had elephants but heavy infantry you would not likely have had such an issue, as the infantry would only have been scratched rather than injured (assuming same damage). Equally, your plan for 28 units in fight 1 may have gone badly had your first battle in the prior province lost units as well anyway. It may pay to have a few extras. You may have needed some extras anyway if you had to fight more than 1 major battle before your 2nd army arrived.

I'm not sure your final comment really applies, realistic? It may not be optimal but depending on how you see the abstractions it may make some sense. Relying on Elephants as your mainstay feels very oddball and high risk, elephants were hardly known for reliability, so I can certainly see it being realistic that your army suffered an utterly catastrophic defeat due to pure bad luck beyond your control. That it happened due to what you see as deployment issues is more a minor detail, and expecting mathematically perfect deployments isn't exactly realistic either.

{mm, took a while to type due to distractions and gray fox said some of the above already)

[Gray Fox]

Like minds...

[lostangelonline]

You were the defender because, as you posted, you arrived before the Seleucids.

Incorrect. From what I posted:

4. Processing the turn, enemy indeed sieges the Ptolemaic region behind, while my troops are advancing towards it:
[..]
6. My best army arrives first to the Ptolemaic region where it engages in battle with the Seleucid army who was sieging the Ptolemaic fortress.

Seleucids arrive first at step 4, while my first army arrives later at step 5. What I meant was that "my best army arrives first" before the weaker army (not before Seleucids who are already there from 3-4 move-steps ago). So I should have been the attacker.

In the second battle, because the elephants are damaged, they end up in the second line.

Even if tired/damaged, if deployed in first row, win chances are higher (considerably if you look at bug 3 when weak cav replace all damaged front-line, which seems to be the same issue). Winning is better than loosing, even if you loose some elephants, so I, as general, would always put those elephants in front. You might say it is realistic because they do not want to fight and the general can not force them, even if that alone will cause loosing the battle; is this what you are saying? (I appreciate you explained to me how the deployment is done by the game code, and I thank you, but I am actually asking why is doing it this way, and not my way or realistically/historically).

The damaged elephants have a ranged attack, and fill into the second support line, removing two archers.

So now that these 2 damaged elephants mess my front front-line, they mess my perfect support-line as well? As a general, if the realistic explication is that I can not convince them to fight in front-line, I put them in 3rd line, never in 2nd. What is the realistic explanation for the general replacing archers with weaker (and damaged) support units?

In the last battle, somehow Anubhavya is in charge, with attack 0. So no attack bonus from the leader.

Oh, 5th bug I missed, thanks. Also don't you agree this is a problem that my better 1-attacking-star general is replaced with weaker 0-attacking-star general?


Anyway thanks Gray Fox for your reply, you have described very well how the game chooses to deploy. I am convinced that I will never choose to deploy like this, and pretty sure it is not even realistic (not all deployment-decisions anyway)

[lostangelonline]

@storeylf: Thank you for your very detailed answer. Sorry for not replying back with same detail, but I am just too tired at 1AM to do it properly, but I will say that, thanks to you (and Gray Fox) I understand much better how game chooses to deploy.

I would have deployed better (and might have won), but some deployment decisions the game make might be realistic (damaged units refuse to fight in the front and general can not force them), while other not (general replaces fresh archers with weaker and damaged elephants, which I consider is bug). I understand that it just happens that final result, due to bug, makes elephants unreliable like in reality, but that is not excuse to have that deployment bug; devs should fix it and come with something else realistic to make elephants unreliable (but they would still be unreliable even if bug is fixed).

Still do not understand why my army is defender, when it arrived 3-4 move-steps after the Seleucids arrived (you say realistic or bug, I am pretty sure is bug), and why my 1-star general is replaced by 0-star general (also sure is bug).

[Gray Fox]

Apologies for mixing up the two battles. I was paging up and down.

You obviously have skills and knowledge. I just believe that you expect too much from Elephants and chariots. Pocus seems to be onboard with adapting the combat line-up system. Good luck!

[storeylf]

I think (but am not 100% sure) the general you get is the one from the larger group in a battle, so maybe your casualties in the 1st battle meant that the +1 guy was now leading the smaller contingent so was not used.

As for the general changing - history is absolutely replete with the best people being sidelined by abysmally hopeless ones, being in charge was not about who was best at the job but about who had the seniority, and that may not have even been miltiary, it was often seniority in the royal family etc. It is perfectly realistic for the 0 rated leader with pampered coward traits to be in charge whilst a+2 uber tactician is sidelined. Indeed in other games (also from slitherine) such a thing is a major feature.

The defender bit I expect is as I said, in game terms arriving 3 or 4 ticks earlier is in essence arriving at the same time (the code has 4 ticks as the cut off point for at least part of the logic). Classing as the defender with no questions asked requires a bit more than having just beaten the other person to the location by few ticks, and there are probably ramifications to changing that in other scenarios.


Realism wise, though:

We can look at history to see some interesting 'deployment' issues - Cunaxa where the Greeks refused to take the center when requested by the general paying them. Crecy where the French were so eager to attack that once they found the deployed English all attempts at control and deploying properly were lost, Cynoscephalae which was a meeting engagment between 2 large armies and there was no chance to pre-deploy at all, Ilipa where Scipio faked the enemy to deploy in a certain way and then came at them with an unexpected formation before they could respond. We can probably continue, but even for these classic key battles deployment wasn't some simple solved issue for each side. Realism and Fate makes sub optimal aspects only to common.

Indeed the game is almost certainly not realistic in many ways, it's an empire building game with abstractions, and you are angling for having the best pre-calculated chance of winning. Nothing wrong with that - it is a game after all, equally I agree that there are probably some glitches that the game could improve upon given how it handles some stuff, but I don't really see what you ask for as somehow 'realistic' per se. In many ways quite the contrary.

Regarding changing stuff. I'm far too familiar with one persons 'fix' being another persons 'breaking change', Indeed I have no doubt some people may complain that your logic would keep sacrificing half dead units they would rather try and preserve to replenish next turn, and they will say such sacrifice is not 'realistic'. One can question whether half damaged elephants should be replaced by archers, but at that point you are getting into specific edge cases rather than easier generalisations, which becomes harder to code and maintain. It is easy for non developers to say 'easy fix' when what they really mean is no more than 'What I want is easy to express', and which is often a million miles away from what is actually easy/economic to do. A younger me always thought wargame AI should be so much better, hence I became a software engineer, now I know it is nothing like as easy as it sounds. I'm sure I could code what you want in this specific case, I'm also sure that without some good knowledge of the existing code and understanding some 'whys' etc that I'd break stuff at the same time.

[loki100]

one small bit, by using two armies, and all the swapping of units between commands, you are chucking away a lot of experience as units and commanders don't match.

worth remembering, if you send two armies into a region and they arrive and fight together, they are 'merged' so one group of units picks up the modifier for having 'joined' a new army that turn.

Not saying this explains everything but from your own description you are reducing the combat capacity of your armies.

[lostangelonline]

Indeed the game is almost certainly not realistic in many ways, it's an empire building game with abstractions, and you are angling for having the best pre-calculated chance of winning. Nothing wrong with that - it is a game after all, equally I agree that there are probably some glitches that the game could improve upon given how it handles some stuff, but I don't really see what you ask for as somehow 'realistic' per se. In many ways quite the contrary.

Thanks for understanding. Indeed this discussion is hard to follow due to it discussing 5 issues at the same time.

Let's re-enumerate the 5 issues, to be easier to track the discussion:
ISSUE 1: army arriving later, is designated Defender (expected: to be Attacker) (realistic explanation: only one army is advancing towards the other?) (game rules/mechanics: if you arrive in a region with enemy army there, you are attacker?)
ISSUE 2: 2 elephants are missing from front-line (expected: to be in front-line) (realistic explanation: damaged units refuse to be in front-line?) (game rules/mechanics: damaged units are never deployed in front-line, except if center can not be filled)
ISSUE 3: weak cav deployed in front-line (expected: more powerful units to be in front-line) (realistic explanation: damaged units refuse to be in front-line?) (game rules/mechanics: damaged units are never deployed in front-line, except if center can not be filled)
ISSUE 4: archers are replaced by weaker support units (expected: always deploy max ranged-attack in 2nd row) (realistic explanation: ????) (game rules/mechanics: max ranged-attack units are deployed in 2nd row, except if large army fills front-line?)
ISSUE 5: 1-star general replaced by 0-star general (expected: 1-star general) (realistic explanation: more powerful/influential royalty/aristocracy always lead lesser ones) (game rules/mechanics: the general with more stars is chosen?)

I think (but am not 100% sure) the general you get is the one from the larger group in a battle, so maybe your casualties in the 1st battle meant that the +1 guy was now leading the smaller contingent so was not used.

As for the general changing - history is absolutely replete with the best people being sidelined by abysmally hopeless ones, being in charge was not about who was best at the job but about who had the seniority, and that may not have even been miltiary, it was often seniority in the royal family etc. It is perfectly realistic for the 0 rated leader with pampered coward traits to be in charge whilst a+2 uber tactician is sidelined. Indeed in other games (also from slitherine) such a thing is a major feature.

(So this is regarding ISSUE 5) Completely agree it is realistic that sometimes you get the better general, and other times the worst general (no matter how the game chooses). You say that when 2 armies meet on the field, the chosen leader is the one leading the biggest army? I thought the game mechanics always chose the best general (from what I've read and experience). Both seem realistic to me, so are acceptable to me, I just want to know which mechanic is used, so I do not make this mistake again in the future. So can someone else confirm which general is chosen in general? Is it always the one leading the biggest army? Or is it random?

The defender bit I expect is as I said, in game terms arriving 3 or 4 ticks earlier is in essence arriving at the same time (the code has 4 ticks as the cut off point for at least part of the logic). Classing as the defender with no questions asked requires a bit more than having just beaten the other person to the location by few ticks, and there are probably ramifications to changing that in other scenarios.

(ISSUE 1) So if two enemy armies arrive in same province in same turn, the attacker will be chosen randomly, no matter who actually arrived first in the move-steps how I thought (which is more realistic). Not ideal, but Is realistic enough, so I can accept that. Thanks. Can someone else confirm (so I can change my future strategies)?

Realism wise, though:

We can look at history to see some interesting 'deployment' issues - Cunaxa where the Greeks refused to take the center when requested by the general paying them.

Regarding changing stuff. I'm far too familiar with one persons 'fix' being another persons 'breaking change', Indeed I have no doubt some people may complain that your logic would keep sacrificing half dead units they would rather try and preserve to replenish next turn, and they will say such sacrifice is not 'realistic'.

(ISSUE 2/3) That would explain why damaged units refuse to be in front-line and/or no general will choose to sacrifice damaged units. Thanks.

Crecy where the French were so eager to attack that once they found the deployed English all attempts at control and deploying properly were lost, Cynoscephalae which was a meeting engagment between 2 large armies and there was no chance to pre-deploy at all, Ilipa where Scipio faked the enemy to deploy in a certain way and then came at them with an unexpected formation before they could respond. We can probably continue, but even for these classic key battles deployment wasn't some simple solved issue for each side. Realism and Fate makes sub optimal aspects only to common.

(ISSUE 4?) I completely agree such situations where deployment is not ideal existed in reality. It would explain the issue in previous post where armies bigger than terrain frontage are deployed wrong when they are attacked, but not when they attack (why attack if you are not ready?). But should this explain why archers are always replaced by weaker and damaged support units? They are not damaged to explain the issues 2 or 3. It still does not make sense to me.

One can question whether half damaged elephants should be replaced by archers, but at that point you are getting into specific edge cases rather than easier generalisations, which becomes harder to code and maintain. It is easy for non developers to say 'easy fix' when what they really mean is no more than 'What I want is easy to express', and which is often a million miles away from what is actually easy/economic to do. A younger me always thought wargame AI should be so much better, hence I became a software engineer, now I know it is nothing like as easy as it sounds. I'm sure I could code what you want in this specific case, I'm also sure that without some good knowledge of the existing code and understanding some 'whys' etc that I'd break stuff at the same time.

As an app and game programmer myself, I completely understand and agree with you here, except that "half damaged elephants should be replaced by archers" is not "specific edge cases", but "easier generalisations". This always happens in many of my battles, no matter the conditions, and is BY FAR the most unrealistic and annoying (because is counter-intuitive and surprising every time) aspect of my battles. I would also argue that this is 'easy fix' compared to all other battle issues presented, since it most clear that it is indeed a bug and needs to be fixed, and to change the code to always deploy, in 2nd row, first the max-range-attack units, and then everything else, seems quite easy to me as programmer, no matter how the deployment code is currently:
-if code first goes through each position and chooses a unit for it, when arrives at 2nd row positions, change to always choose best range-attack unit first;
-if code first goes through each unit and chooses a position for it (this is more hard), when arrives at archers and they have bigger range-attack than other unit already in 2nd row, change to replace that unit (which is reassigned to other position).

[lostangelonline]

one small bit, by using two armies, and all the swapping of units between commands, you are chucking away a lot of experience as units and commanders don't match.

worth remembering, if you send two armies into a region and they arrive and fight together, they are 'merged' so one group of units picks up the modifier for having 'joined' a new army that turn.

Not saying this explains everything but from your own description you are reducing the combat capacity of your armies.

Excellent point. I missed that mechanic when 2 armies fight together. So in the last battle, all elephants+HCav+archers lost 1-star worth of experience (which in battle calculations is -1 to the unit total, right?). This makes it even more important to know which general will command if 2 armies fight together (as you might actually loose a battle you win with an army, if you bring a 2nd army to help, which seems unrealistic). Thanks.

[storeylf]

As an app and game programmer myself, I completely understand and agree with you here, except that "half damaged elephants should be replaced by archers" is not "specific edge cases", but "easier generalisations". This always happens in many of my battles, no matter the conditions, and is BY FAR the most unrealistic and annoying (because is counter-intuitive and surprising every time) aspect of my battles. I would also argue that this is 'easy fix' compared to all other battle issues presented, since it most clear that it is indeed a bug and needs to be fixed, and to change the code to always deploy, in 2nd row, first the max-range-attack units, and then everything else, seems quite easy to me as programmer, no matter how the deployment code is currently:
-if code first goes through each position and chooses a unit for it, when arrives at 2nd row positions, change to always choose best range-attack unit first;
-if code first goes through each unit and chooses a position for it (this is more hard), when arrives at archers and they have bigger range-attack than other unit already in 2nd row, change to replace that unit (which is reassigned to other position).

I agree that it would not be so bad if kept to such generic concepts (order by high range attack, deploy as support) etc. I was meaning that getting down into if elephant and if archer etc, that is less generic and heads much more into special case handling.

The problem with generic logic is what you have now, it tends to be not so good at certain cases. The logic they have now is fairly generic and you have the current problem. I wouldn't be surprised if doing what you suggest would create some other case that doesn't exist now and then was seen as a bug.

That's not to say I disagree with what you want, I'm sort of neutral on how it does it. I don't mind the lack of certainty as I don't expect predictable battles etc, so I don't see it as a clear bug per se. At the same time I don't have a problem if it does what you want either.

There is no correct answer to the best auto deployment. I believe archers in support carries more risk to them than reserve (flanking and pursuit damage), so isn't pure benefit. I seem to think I saw some script that reduced damage to light cav, which if true would mean having light cav at the front may actually be good in some cases - it may help you get away with more of your army if the battle looks bad.

If this was a board game for example I wouldn't follow the same pattern every battle. In your situation I would of course be free to deploy injured elephants forwards, chariots on flanks and archers behind if I was sure that gave me the only chance and I didn't mind losing the half dead elephants even if I won. However, in another battle I may well decide that a different combo would be just as good and I could keep my experienced injured elephants safer for a follow on battle I could see coming or just to not lose them, I may even feel that the best archers should be kept safer in reserve for the next battle if I thought they were overkill now or that I would lose anyway and reserves are where I had most chance of keeping stuff alive. Many things feed into such decisions, and it is hard to do the auto deploy that satisfies everyone.

Whilst me may prefer it if it wasn't, how the code is currently written is almost always important no matter how easy you can define your logic on paper. You have to change it in a way that doesn't break stuff you hadn't thought of. I was changing the defender choice logic the other day, and hit a few snags - not realising that for example some groupIds had been set to null before the full battle UI logic was over with, so kept getting errors towards the end of the battle as it was still calling my logic. Or some wrong assumptions about how the leaders ratings are setup if there is no leader, and then had a timing issue with part of the battle logic wanting to know who was the defender before I had the info i wanted to use to work out the defender, so had to then reorder other code which in turn risked creating other timing/order bugs etc. It was easy to define on paper what I wanted, getting it to work couldn't ignore the existing code and logic flows.

[loki100]

one small bit, by using two armies, and all the swapping of units between commands, you are chucking away a lot of experience as units and commanders don't match.

... .

... This makes it even more important to know which general will command if 2 armies fight together (as you might actually loose a battle you win with an army, if you bring a 2nd army to help, which seems unrealistic). Thanks.

given how Pocus usually codes stuff my guess would be the largest army is the one that usually retains the leader (& I'd guess here that largest is defined by units) but there will be some sort of random element just to keep us all on our toes

I think it sort of encourages a 'hurry up and wait' model of moving to a decisive battle. So you see something like the Second Macedonian War with a couple of years of the two armies staring manfully at each other [1] and then a decisive clash - or in our terms, the Roman army stands still while new units are fed in and assimilated then moves in for the kill?

[1] I do realise there was more than manful staring going on, but neither side seemed to be looking to force a major clash early on.

[Pocus]

Seleucids arrive first at step 4, while my first army arrives later at step 5. What I meant was that "my best army arrives first" before the weaker army (not before Seleucids who are already there from 3-4 move-steps ago). So I should have been the attacker.

A group who arrived recently (4 or less impulses ago ) is considered "not settled in" and is still moving through the region, so does not qualify imperatively as defender. In case of tie, the largest army will defend. Think of that as a meeting engagement in fact. Nobody was settled in the region and both armies were still moving through the region.

Hopefully the improved deployed algorithm (who has 2.5 more tests and conditions than the previous one) will do a better job in the next patch.

[lostangelonline]

A group who arrived recently (4 or less impulses ago ) is considered "not settled in" and is still moving through the region, so does not qualify imperatively as defender. In case of tie, the largest army will defend. Think of that as a meeting engagement in fact. Nobody was settled in the region and both armies were still moving through the region.

That's quite a very good and realistic explanation that makes perfect sense and is intuitive. Thanks!

Hopefully the improved deployed algorithm (who has 2.5 more tests and conditions than the previous one) will do a better job in the next patch.

Really glad battles have been improved; will wait for the patch then.

PS: also big thanks to storeylf for taking the time to discuss in detail, I agree with you and understand (did not know you are one of the devs). Also thanks to loki100, Gray Fox, and everyone for all the answers!

[storeylf]

PS: also big thanks to storeylf for taking the time to discuss in detail, I agree with you and understand (did not know you are one of the devs).

I'm not one of the devs , just to be clear. Sorry if you got that impression. If it was due to me talking about changing the script, that is because it is part of the install and you can play with and mod it as you want.