Military Model VI

[Mark_Everson]

Re: Manpower

Originally posted by LDiCesare
2)At the end of each round of fight, check per unit the number of casualties. If the unit is below a threshold, it tries to flee.
Odds to manage fleeing are always less than 100%, I don't change that.
The problem with 2) is that, right now units don't heal. Thus, a unit which fled once will always flee. (snip)

Hi Laurent, sorry its taken me so long to comment on this, hope its not too late. I think a simple solution to the problem above is to make the morale check based on the percentage of the unit lost in the Current battle. So say a morale-based retreat starts when strength is reduced to 70%. That check is triggered when the unit's strengths is 70% of what it was At the Battle's Start. We would of course have to carefully define what is a distinct battle and what isn't. It seems to me this gets over the problem of perma-fleeing units, and really makes more sense anyway in terms of the real world.

[LDiCesare]

That may be done. I will try it. We should figure what percentage of losses are allowed per element kind. I will try to code it and check how many more rounds fights to death would need with a 50% threshold, then adjust in order to retain correct gameplay.
Any better suggestion that would avoid my adding a start of fight (turn) health would be welcome, though.

[Mark_Everson]

Combat system determinism

This issue came out of an email discussion that I thought should continue here.

I have to admit the combat results in D7 seem always a bit too determined by odds for my tastes. In history its not nearly so predictable. I favor a bit more uncertainty. IMO something like one time out of 20 even an army outnumbered 2:1 (given everything else equal) should be able to win a battle.

But we Definitely don't want uncertainty at the ridiculous Phalanx-beats-Tank level seen in Some games .

[LDiCesare]

The big thing I am afraid of in the combat model is there may be a too big advantage for defenders due to the change of element modelization. The fact that figures are more important (instead of a 10 health unit we have a 5000 health unit) probably show things that were not apparent before. I am not so much afraid of randomness as of behaviour which would differ from the previous code in a strange way.

This may be the time to ask what the priority should be between:
Naval combat - I am totally useless in terms of modelization here as I know very little about naval warfare- Gary, ideas?
AI
Population taking arms to defend itself against a conqueror - this means I need to know in the population model the part of population that is available for fight. Determining the proportion, reason why an EG would try to defend itself against an invader being left for later. This one was a D6 remark that cities fell too easily.
Siege warfare.
Other stuff?

[Mark_Everson]

Originally posted by LDiCesare
The big thing I am afraid of in the combat model is there may be a too big advantage for defenders due to the change of element modelization. The fact that figures are more important (instead of a 10 health unit we have a 5000 health unit) probably show things that were not apparent before. I am not so much afraid of randomness as of behaviour which would differ from the previous code in a strange way.

Hi Laurent:

Yeah, we definitely need to know if the model has been effectively changed after the refactoring.

This may be the time to ask what the priority should be between:
Naval combat - I am totally useless in terms of modelization here as I know very little about naval warfare- Gary, ideas?
AI
Population taking arms to defend itself against a conqueror - this means I need to know in the population model the part of population that is available for fight. Determining the proportion, reason why an EG would try to defend itself against an invader being left for later. This one was a D6 remark that cities fell too easily.
Siege warfare.
Other stuff?

I can start to work on a naval model, but it wouldn't be for a while. Too many other things I'm behind on already! I guess either Gary, or maybe a newbie I can get in by advertising?

My take on the order (including difficulty) would be:
1. Population fights
2. Siege Warfare (since at least a start can be made in a straightforward manner, more detailed modeling would follow.)
3. Naval rules
4. AI

[Gary Thomas]

I will think about naval combat, and post something soon.

Cheers

[Mark_Everson]

I was about to ask "Whatever happened on the idea of disbanding military units (or TFs)" but uncharacteristically thought I should see if it was in already. It was, and works great! Thanks for doing it Gary! My only quibbles at this point are:

1) that the map and detail frame need to be updated after the operation is complete.

2) for squares that currently have no population the order kills the unit but puts no population down. we had some discussion of this via email, but it wasn't clear to me what you meant about "using manpower but getting 0" (I'm paraphrasing). I still would prefer to have units on virgin territory start a new economy a la colonization. I can do it if you like.

Cya,

Mark

[Gary Thomas]

1) that the map and detail frame need to be updated after the operation is complete.

feel free to fix this.

2) for squares that currently have no population the order kills the unit but puts no population down. we had some discussion of this via email, but it wasn't clear to me what you meant about "using manpower but getting 0" (I'm paraphrasing). I still would prefer to have units on virgin territory start a new economy a la colonization. I can do it if you like.

The unit would have no women, so the disbanded soldiers will wander off somewhere else looking for them, rather than start a male only colony.

Cheers

[Mark_Everson]

Originally posted by Gary Thomas

Hiya Gary. The unit would have no women, so the disbanded soldiers will wander off somewhere else looking for them, rather than start a male only colony.

Already thought of this. Because of support personnel and "camp followers" there should be something of the order of enough women around to form a colony. (We really should remove more people from the economy than the raw fighting strenght to include the "tail" of the army. Also, I just think its simpler for the player and avoids micromanagement. If the player wants a colony Here and is willing to pay a military unit to get it, I say give it to them. If it turns out settling with military units leads to bad Gameplay then by all means we shouldn't allow it. But we shouldn't allow relatively unimportant details to stand in the way of fun IMO. YMMV.

[S. Kroeze]

Dear Mark_Everson and others,

Mark recently suggested that I could perhaps be of use while helping you to create a new Military Model.
Though military history is definitely NOT my subject I am willing to try to collect useful data.

I copied what follows from an ancient post, once made by myself on the GGS Forum.
I hope it is useful; if not, please say so!
For the moment I do not have the energy to invent an Attrition Model myself. Yet I guess this post shows rather convincingly how quickly military forces disintegrated.

I also would like to make some remarks about warfare in general. Most games -and Civilization is no exception- present a completely distorted picture. The units in CivII seem to possess some 'divine' immortality. In reality most military campaigns were relatively short and armies disintegrated as quickly as they were recruited.

'Disease was a greater threat to the health of Civil War soldiers than enemy weapons. This had been true of every army in history. Civil War armies actually suffered comparatively less disease mortality than any previous army. While two Union or Confederate soldiers died of diseases for each one killed in combat, the ratio for British soldiers in the Napoleonic and Crimean wars had been eight to one and four to one. For the American army in the Mexican War it had been seven to one. Only by twentieth-century standards was Civil War disease mortality high. Nevertheless, despite improvements over previous wars in this respect, disease was a crippling factor in Civil War military operations. At any given time a substantial proportion of men in a regiment might be on the sicklist. Disease reduced the size of most regiments from their initial complement of a thousand men to about half that number before the regiment ever went to battle.(!)

Sickness hit soldiers hardest in their first year. The crowding together of thousands of men from various backgrounds into a new and highly contagious disease environment had predictable results. Men (especially those from rural areas) who had never before been exposed to measles, mumps, or tonsilitis promptly came down with these childhood maladies. Though rarely fatal, these illnesses could cripple units for weeks at a time. More deadly were smallpox and erysipelas, which went through some rural regiments like a scythe. If soldiers recovered from these diseases and remained for some time at the training or base camp -where by poor sanitary practices and exposure to changeable weather they fouled their water supply, created fertile breeding grounds for bacteria, and became susceptible to deadly viruses- many of them contracted one of the three principal killer diseases of the war: diarrhea/dysentery, typhoid, or pneumonia. As they marched southward in summer campaign, many of them caught the fourth most prevalent mortal disease: malaria. A good many Union occupation troops in southern cities as well as Confederate soldiers camped near other cities -especially Richmond- experienced another soldiers' malady, venereal disease, of which there were about as many reported cases as of measles, mumps, and tonsilitis combined.

Disease disrupted several military operations. Lee's West Virginia campaigns of 1861 failed in part because illness incapacitated so many of his men. One reason for the abandonment of the first effort to capture Vicksburg in July 1862 was the sickness of more than half of the Union soldiers and sailors there. Beauregard's decision to abandon Corinth was influenced by illness of epidemic proportions that put more than a third of his army on the sicklist. By the time Halleck's Union army had established its occupation of Corinth in early June, a third or more of the Yankee soldiers were also ill. Nearly half of the twenty-nine Union generals came down sick during the Corinth campaign and its aftermath, including Halleck himself and John Pope with what they ruefully called the "Evacuation of Corinth" (diarrhea) and Sherman with malaria. Halleck's failure after Corinth to continue his invasion into Mississippi resulted in part from fears of even greater disease morbidity among unacclimated northern soldiers in a Deep-South summer campaign.'
(source: J.M.McPherson: 'Battle Cry of Freedom',1988)

Here are some hard figures about the Thirty Years' War(1618-1648), underlining this same point:

'One might wonder why any man would freely join such a force; and indeed, many soldiers served in the ranks against their will. The troops from Sweden and Finland, for example, were recruited by a form of conscription known as the indelningsverk, which obliged a specified community to provide a certain number of soldiers. Most of them were peasants: in the voluminous (but as yet little analysed) records of the Swedish and Finnish forces serving Gustavus Adolphus and his daughter, bönde (peasant farmer) is by far the commonest entry in the enrolment lists. They came from villages like Bygdeå in northern Sweden, which provided 230 young men for service in Poland and Germany between 1621 and 1639, and saw 215 of them die there, while a further five returned home crippled. Enlistment was thus virtually a sentence of death and its demographic impact was profound. The number of adult males in Bygdeå parish steadily decreased -from 468 in 1621 to 288 in 1639- and the age of the conscripts gradually fell as more and more teenagers were taken, never to return. The social impact was also high: at first, the 'idle poor' tended to furnish most of the recruits, but after a while it became the turn of the younger sons of more prosperous families, and finally the only sons of even rich peasants were called to die away in Germany. In some smaller settlements, by the end of the 1630s, every available adult male was either on the conscription lists, already in the ranks, or too crippled too serve. Total losses in the Swedish army between 1621 and 1632 have been estimated at 50,000 to 55,000; those between 1633 and the war's end were probably twice as high. Clearly the war was causing depopulation in Sweden and Finland on an unprecedented and -ultimately- unbearable scale.
(NB.: Since the population of Sweden in 1600 is estimated at only well over one million these were unbearable losses indeed; Finland had only about 200,000 inhabitants)

Of course, there were many other causes of military losses unconnected with fighting. When Christian IV's headquarters were at Tangermünde on the Elbe, in 1625, "the stink of the camp got up one's nose" (in the words of a chronicler) and, before long, disease had reduced the Danish forces substantially. The Imperialists quartered in Hesse-Darmstadt during the winter of 1634-35, after their victory at Nördlingen, were forced to sleep ten and twenty to a house; it was therefore not long before illnesses due to overcrowding took their toll. In the Scots Brigade serving in Germany between 1626 and 1633, some 10 per cent of the regiments were sick at any one moment, with epidemics increasing the rate dramatically from time to time. For example, the Scots who garrisoned the lower Oder in 1631 lost 200 men a week from plague, and more still from camp fever (typhus) and the other illnesses common among early modern armies.'

Wastage rates in selected regiments:
4 English regiments:
1627 June: 4,913
1627 Oct.: 3,764
1628 Apr.: 1,882
1628 May : 1,630

4 Scots regiments:
1630 Jan.: 1,900
1632 Mar.: 1,300
1634 Oct.: 200

3 Swedish regiments:
1631 Sept: 2,577
1632 Mar.: 1,212
1632 Dec.: 828
(source: G.Parker: 'The Thirty Years' War',1997)

Parker arrives at a monthly loss of lives ranging from 2 to 20%.
And here is an example of an army that almost literally melted away:

'Meanwhile, Napoleon's Grande Armée was shambling across the Russian plain. It was a long, hot summer and thousands of soldiers succumbed to heat exhaustion before it was decided to march only at night. Discipline had degenerated. Supplies were short. Evidently, Napoleon's quartermasters had counted on scavenging to meet a large part of their daily requirements, but the retreating Russians left behind them little that was edible. Most successes that were achieved in foraging were made by the more experienced French soldiers and this, added to the stingy paternalism of the French quartermasters when distributing supplies to the contingents of their allies, intensified the hostility of the non-French towards the French. The horses suffered even more from the supply shortage; many of the died from bad feeding, while others were themselves eaten by the soldiers. By the end of August many of the non-French troops were barefoot and their dust-caked uniforms little better that rags. Stragglers were numbered in the thousands; some of these were merely ill, for typhus and dysentery were spreading, but the majority were lagging behind voluntarily as a tactical preliminary to desertion and return home. Above all, water was short. The few good sources were usually tainted; occasionally a corpse or an amputated limb would be found in a spring or pond, deposited by the retreating Russians and perhaps recalling to the French soldiers Napoleon's utterance at Smolensk: "how sweet smells the corpse of an enemy!"

Thus, despite the legend of the subsequent retreat from Moscow, it was the advance which caused most damage to the invaders. There was never any need for the Russian army to fight a set-piece battle, for Napoleon's forces were disintegrating day by day. However Kutuzov was persuaded by public opinion, pressure from St.Petersburg, and the enthusiasm of his junior officers, to make one stand before Moscow. So in early September the Russian army stopped retreating and formed up in a defensive position on high grounds near the village of Borodino, commanding the Smolensk-Moscow highway about seventy miles west of Moscow. The Russian forces numbered about 120,000, of whom 10,000 were hastly raised and half-trained militia sent out from Moscow.

By this time the French army had shrunk from the half-million, with which it had begun the campaign, to a mere 130,000, and was already slightly inferior to the Russians in artillery. Napoleon decided to make a frontal attack on the opposing positions, probably fearing that otherwise the Russians would be tempted to retreat once more and deny him his long-desired victory. At the end of a hard-fought day, in which positions changed hands time after time, the Russians began a slow and orderly withdrawal. Although the French had won the battlefield they had not won the war, for they had not destroyed the opposing army. At Borodino, which Napoleon later adjudged his most expensive and terrible battle, the French suffered 30,000 casualties and the Russians 40,000 (some Russian historians give the French casualties as 68,000; some French historians give the Russian casualties as 60,000). Bagration was mortally wounded and died some weeks later at his country estate. Barclay, as though conscious of his recent unpopularity, deliberately exposed himself in the battle, emerging a hero and uninjured.

A week after leaving Moscow there was a heavy engagement with Kutuzov, resulting in both sides retiring to lick their wounds. Napoleon was forced to retreat along the road by which he had advanced, via Borodino with its still-unburied corpses. Although Kutuzov, to the disgust of the British general who was attached to his headquarters, avoided close engagement with his retiring opponent, Cossacks and partisans were quick to bring a bestial death to stragglers or lagging detachments. The first snow fell in the first week of November. Neither the cavalry nor the artillery had been supplied with winter horseshoes and it was not long before most of the army's horses disappeared. Without horse transport the supply situation became catastrophic and more men dropped out by the roadside. At the end of November the Russians were outmanoeuvred for a few precious hours, enabling most of the army to cross the river Berezina, but thousands of stragglers and camp followers were killed at this point, drowned, trampled underfoot by their comrades, or massacred by Russian guns. It was not until early December that the really cold weather set in. Contrary to the impression given by Napoleon's apologists, there was no early winter that year and the French were not defeated by the cold. The cold, when it came, only finished the job. In the final count probably four-fifths of the half million men who followed Napoleon into Russia were lost, and only a few of these died from cold.'
(source: J.N.Westwood: 'Endurance and Endeavour, Russian History 1812-1992',1993)

So in the end Napoleon -who is generally considered a fairly good general- was defeated by an efficient use of scorched earth policy, inadequate supplies, weather conditions (summer heat and winter cold), diseases and mass desertions.

Finally I would like to remark that it seems to me rather premature to condemn a game system not yet existing as inevitably boring, because it could incude an element one doesn't like, whether preplanned turns or realistic delay of information/command dissemination. By far the most exciting and demanding game I have ever played is the board game Diplomacy, which has preplanned turns, but no information lag. Yet I do not doubt there also exists some boring game with preplanned turns! In the end tastes differ.

Please tell me whether this information is useful or not!
And Mark: perhaps you can tell me in more detail which questions you need to have answered.

Thank you!

S.Kroeze

[Gary Thomas]

A very good and very useful contribution!

There are quite a number of valid points regarding army maintenance included here.

Unfortunately, from my point of view, the deficiencies of the miltary model are somewhat more fundamental than the issues addressed here.

The following comments relate to the present code in Clash which is restricted to ancient era warfare. Generally this can be taken to be pre-gunpowder warfare. The following comments are addressed to that situation.

The move time in Clash is about 5 years, with a rather ill-defined "miltary turn length" of a month. Nevertheless, successive moves are five years apart in game time.

There are very few armies in that era that lasted for five years. A couple of exceptions are Hannibal's army and those of the Mongols. Even they would have had virtually a complete replacement of personnel in that time.

This, effectively, means that armies should be built, have a single move, then be removed by attrition.

Unfortunately movement is designed around the military turn, and, in that time, the armies normally cannot reach any reasonable objective. So the combination of turn lengths effectively makes realistic simulation of the military situation intrinsically impossible.

Since a realistic model is impossible, we are stuck with an unrealistic model.

But we have that already...

Cheers

[LDiCesare]

Maybe putting an upkeep cost in men would help a bit. If we consider 1 month to be 5% damage to an army and ask a corresponding upkeep cost in men, one military turn would still deal 5% damage to the supporting population. Alternatively, I could deal random 2-20% damage to every unit at the end of every turn (adjusted by whatever techs we would have) and let the player heal their units. This is better as far away armies may not be healed (?) and you could probably better control the healing/upkeep cost. This requires healing. I don't know why I haven't coded it yet... Probably I don't want to see the econ part of it.

[Mark_Everson]

Hi S. Kroeze, thanks for the congtribution! For the record, here's what I was asking for on the GGS forum.

...we need someone to work on a model of naval transport and warfare. We have ships in the game (they will be more prominent in the next demo) but its at the level of the cheezy civ model of naval transport. Men can be moved across the world in triremes with no attrition, etc. If you would like to summarize what you think are the most important real-world effects that we need to get right, it would be of significant value. If you were interested in roughing out a game model for it, it would be even better!

We are in a similar position on city-related issues like sieges, and that would be an alternative valuable thing for you to organize facts on.

Since then, Gary has said he will come up with a naval model. But still thoughts of the sort your cited for attrition on the topic of naval transport, combat, and siege warfare would be quite useful. We're all fairly conversant with military history, but that's not to say we can't miss something fairly important! One of the things I hope to capture is that invasions using naval transport from long distances away are almost an impossibility. Disembarking at a nearby port should generally be required. Certainly before fairly recent history anyway.

Laurent, I really like your idea of an upkeep cost in men. Of course the attrition rate should depend on many factors. I am a bit worried that players won't like armies evaporating on them despite historical precedent! I think we can work out something of the sort you envision without huge problems. But I'd prefer to not have to deal with it for a bit . (I expect you're of that mind too, given your comment on the potential nightmare it'd be with the economy.) A few demos from now I think we should put in something like it though.

[LDiCesare]

Note that disease is most important for sieges. I can't see how we can model besiegers and besieged without taking disease into account. Besieged would starve and die (both the army and the population), while besieging armies would die of disease. Peasants in the countryside would have long fled to the other end of the square, so they may not need be impacted.

Here are the current military code things to do:
0. Tweaking existing code so costs/fight outcomes are to our liking(mostly mine as I run many simulations to get odds correct ).
1. Population fights -> Need to touch the population model to have some demography in it. I'll either put a male female population figure in the demography classes or do something more or less clean.
2. Siege Warfare -> Need some rules for dying of disease. Does a square with no food have people in it dying, and can I retrieve that information to propagate the damage to units in the city?
3. Naval rules
4. AI
5. Healing units. I'd probably push that one up around 2 if I feel it is needed.

[LDiCesare]

Elements in Units

I describe quickly how elements and units are now modelled, and the schmonsequences on some areas of the model, mainly the battle itself.
Units are made of several elements. Each element is made of a number of men. There cannot be two elements of the same kind in a unit.
For instance, the old model with 500-man elements had:
Phalanx unit = 2 skirmisher elements, 3 light spear elements, 5 heavy spear elements.
The new model is:
Phalanx unit = 1 skirmisher element worth 1000 men, 1 light spear element worth 1500 men, 1 heavy spear element worth 2500 men.

How battles were processed:
Elements were lined up one against another. This was divided in left and right wing, support and reserves.
The front lines had to have the same "frontage", i.e. the same number of men, i.e. the same number of elements. If there were not enough elements to make up a front line, elements which would have been in support, like archers, had to fight at the front. For instance, an army of 10 attacking warriors against defenders made of 8 warriors and 4 archers would behave this way:
Ideally, all warriors fight in the front line, and archers in support. There are two attacking warriors too much however. So, two archers go in front line. Thus we have a defending deployment of 8 warriors and 2 archer elements in front, 2 archers in support.

With the new model, I have 1 attacking element of 5000 men, 2 defending elements, warrior for 4000 men and archer for 2000 men. I can no longer use the 1 element for 1 element code, because I would get a different result, namely that all 2000 archers fight in front line instead of only half of them.
One could say that splitting the 2000 in two is the obvious solution, and I just need to find a way to code it.
Maybe.
But:

That means if you have 4001 warriors against 4000 + 1000 archers, I would have to move either 1000 archers or 1 archer to front row. This is silly as it looks much better to have no archer or a small company sent to help fighting in melee while the rest fight as real archers.

So the question is: Do I model individuals who fight by themselves or do we try to use something like tactical units (aka elements) as atoms in fights? Or is there a better way (using thresholds, or randomness?) to model whether an element supposed to be in support should help the front row, and how many men it sends. The way I present it is biased, as it is not really a choice, considering the opponent outflanks the first line of defense. Still, I don't have any brilliant idea of how to model this.

The main drawback of the current code is that there will be arbitrary scale differences between a fight with few elements and a fight with many elements, only because of the modelisation itself, which is an artefact.