Military Model IV

[LDiCesare]

Supposing I have 2 units made of 4 supporting elements and 6 assault elements, and one is assigned as support unit and one as assault unit, against one unit of 10 nondescript elements, here is the matching I'd like to have (I suppose here the assault elements can handle only 1 nondescript without needing reinforcements):
1 unit in support (all elements),4 supporting elements in and 6 assault elements matched against 1 element each.
That is a recursive approach in terms of code, which means it is small code, fast to write but maybe not fast to execute. In terms of simulation, I prefer this to a (8 supporting elements, 12 assault elements 2 vs 1 twice, and 1 vs 1 8 times) approach. It is less efficient but it represents better the fact that during war, you stick with units of your battalion, sometimes the next one, but a single battalion is not very often broken up into lots of different locations because they would be more efficient there: it would be very hard in terms of logistics and command line to reorganize an army that efficiently. (From my time in the army, I believe it would be really hard for officers).

[LDiCesare]

I finally coded the model wholly (well, except for tons of stuff like leaders, long range and so on...) so it seems to work fine.
A few questions: How many rounds of fight should there be?
There is no round yet at scouting phase because I have a code that is good only for archaic units. There can be up to 10 rounds, more probably 3 rounds, during manoeuvre phase. How many rounds in assault phase?
Here are a few figures to give an idea of the mortality:

2 Hordes vs 1 Phalanx: 7 rounds: 4% chance of phalanx dying. 9 rounds: 16%. 10 rounds: 28% 12 rounds: 54%
1 Phalanx vs 1 Phalanx: 6 rounds: 2% chance of 1 dying. 7 rounds: 5%. 9 rounds: 16%. 12 rounds: 30%.
1 Light cavalry vs 1 phalanx in 12 rounds gives 30% vs 13% in favor of the cavalry.
Note these are non-experienced, non-trained troops with no leader. Also, defenders don't use fortifications yet.

Casualties begin to appear after a number of rounds roughly equal to the number of elements in the units.
Now, anyone has ideas of figures for bonuses of terrain and foritifications?

[Mark_Everson]

Excellent Laurent! That's great progress .

Now the hard work of playtesting the model starts! I hope Krenske he will have some time available soon... Do you have methods to provide detailed output of what happens during a battle? I haven't looked at the test methods you sent me, so they may be in there already. But if you don't have those yet, I think that is the next step. That will allow playtesters to see what happened in detail, and fix any unforeseen problems.

The results you show are somewhat surprising to me, in that I would really expect a single phalanx to win against two hordes. Greek phalanxes very frequently won battles in which they were seriously outnumbered against the Persians, because of the many advantages of the phalanx shock attack. I would also expect a phalanx to win vs. light cavalry unless the light cavalry had bows and refused to close with the phalanx. Also, I want to dispel this idea that the battle should always go to the death of a unit or TF. Unless one side manages a complete encirclement as at Cannae, the losers will always retreat well before they are annihilated IMO. I seem to recall there are morale checks somewhere in the model, do you have those working yet?

I could make guesses about bonuses for terrain fortifications and defense, but would prefer to wait for Krenske to take a shot at it.

It's great seeing this come together. Thanks for all the hard work!

[This message has been edited by Mark_Everson (edited March 18, 2001).]

[Richard Bruns]

Thanks, LDiCesare!

I'd love to playtest the model. I also need to see how different units compare so I can start to set up the military side of the tech tree.

[LDiCesare]

Here is a sample output that I get from one round of fight. Note all the messages are passed by observers so you can easily store them in file, scrap them or pop them in a window if you like. There's really a lot of details I can output (I can put more than what's here). Does anyone have a clear idea of what output they would like to see? I don't want to output a 1Mo file for a fight between two units.
I think we could use two outputs, one for debugging/twitching the model, and one, less detailed, to give reports to the player about important battles.

Meeting of armies of civs Civ1, Civ2 in square game.data.MapSquare@6ed786
Army:
Attacker of civ Civ1:Att Horde0 of civ Civ1	Att Horde1 of civ Civ1	Att Horde2 of civ Civ1	
Army:
Defender of civ Civ2 [img]http://apolyton.net/forums/biggrin.gif[/img]ef Phalanx3 of civ Civ2	Def Phalanx4 of civ Civ2	Def Phalanx5 of civ Civ2	
Scout output phase: 0.0 in favor of Civ2
Fire support for Civ1: 0.0
Fire support for Civ2: 0.0
Fire support for Civ1: 0.0
Fire support for Civ2: 0.0
Fire support for Civ1: 0.0
Fire support for Civ2: 0.0
Manoeuver output phase: 0.13484573 in favor of Civ2
End of manoeuvre: Status: 
Attacker of civ Civ1 : 25.0 health	
Defender of civ Civ2 : 22.0 health	

Fire support for Civ1: 0.0
Fire support for Civ2: 0.0
End of combat: Status: 
Attacker of civ Civ1 : 23.0 health	
Defender of civ Civ2 : 20.0 health

I corrected a few flaws in the code which ran against the phalanx, but the results are still not good enough. I am going to try and change a few figures to get it right. Without any more input from Krenske, here's what I choose: I'll take 10 rounds of fight in the assault phase, and may increase the health of each element. Currently, 1 element = 1 life = 1 death when a hit is scored (cf assault model).

[Mark_Everson]

Hi Laurent:

It looks good to me, aside from the bloodless result, which you are working on .

I think for testing, the output needs to be a bit more detailed. You could write it out to the temporary output IO.out I've set up, or to a file as you said. What we want the player to see will be the result of a lot of testing and refinement, so I'd say just take a quick stab at that, and playtesters will tell you what they really want to see. But we should at a minimum see who remains in posession of the battlefield, losses of men, etc. In the long run we were thinking to put out a narration of the battle (4-5 sentences) that could give the player info on why they won or lost.

-Mark

[LDiCesare]

Here is my take on the model:
It seems to me the following figures apply: Given the same defence, the same health, an army has to have 4 times its opponent's offensive power in order to beat them when outnumbered 2 to 1. Here is why:
(Krenske, please correct if I misread you)
Each side deals Power attacks, which, all other things being equal, deals Power damage. These are divided (more or less evenly).
So if you have side1 with 1 unit, power = 4, and side 2 with 2 units power = 1, side1 deals 4 damage and suffers 2, but side 2 has twice as much life. Thus it is a match.
That's why the phalanx example with 70 attack against a warrior of 20 currently loses 2 to 1.
Do you think players and mod makers will understand that, or do I scale somehow the figures (square, whatever) in order to allow a 2 to 1 ratio to be withstood by an army with only double firepower?
Basically, 2 to 1 is currently 4 times firepower, 3 to 1 is 9 times and so on. This means superior numbers are a very definite advantage.
The squaring would mean deadlier fights given the current figures: Currently, the 1 phalanx / 2 hordes has 18 attacks vs 10 + 10 attacks (with a 10% hit, means 1.8 vs 2 damage sustained).
Squaring the attack and defense (not changing armor), this leads to: 62 attacks instead of 18, and still 20 attacks on the other side. That way, keeping the current unit stats, it would require 3 warriors to outdo one phalanx (because the phalanx loses firepower more reapidly as it is hurt).

Now of course, I may have totally misunderstood the model. ;P
[This message has been edited by LDiCesare (edited March 27, 2001).]

[Mark_Everson]

Hi Laurent:

What you say sounds reasonable in the absence of morale. But what would happen in the real world (at least on open ground) is that the phalanx would crush one of the hordes, its broken remains would flee, and the other one would likely decide it doesn't want to be there either.
So to see what morale is supposed to do, I looked at Krenske's model page. I'm completely mystified about the morale model, since it does things that don't match reality in any way. I suspect there's some typos or other errors in the formula:

Morale test% = (((morale + experience + training + terrain mod) x 5) - 1 per hit below established strength) x multiplier effect

Since given what the table says morale will decay to a low value quickly, and moreso if a unit is winning ?!? (In table high numbers are worse for morale).

Is there a Big misunderstanding I have? Is that the way you read it too?
I'm going to email Krenske and see if I can pull him into this conversation.

[Richard Bruns]

A question about Task Forces:

In old threads from 1999, I read about the possibility of "Task Force Workshops" that allowed players to customize their army groups by adding to gether the basic units. They combine the available units of infantry, cavalry, archers, and auxiliaries to create a special army type. Players could experiment with these, trying to make the most cost effective unit or the one that fits their style best. Once the player creates a TF template, it becomes a production option in the Economy model.

Are we still doing this?

Another option is to make TF composition depend on technologies. For example, the game default Phalanx might be a task force of 3 Hoplite units. This technology gives a 15% power bonus to the three units organized in this manner, thanks to the new tactics developed that allow this military organization ot be more than the sun of its parts.

If we did this, we would have to abandon the TF workshop idea or restrict it. If players automatically get the bonus fo combining the 3 units themselves, the technology is useless. We could set things so that combining the 3 units before the Phalanx application is researched wouldn't get the benefit, but that might make the workshop pointless.

What is the current plan for creating task forces?

[Mark_Everson]

How things are now (AFAIK ):

Elements are the fundamental building block of the system. See Krenske's post of 2/15 for some sample element and unit designations. Elements are (will be) Applications in the tech model.

Units are built up of elements. We will have standard units. Players will be able to specify non-standard units, composed of any elements they can build, in a Unit Workshop. There will of course be some limitations on how may elements you can cram into a unit depending on technology etc. You can't swap around elements between units in an arbitrary fashion (I think...), but you may be able to fuse two badly beaten up units.

TFs are composed of units. TFs are what the player moves, although TFs can be disolved into their component units and reshufled in the field just as in demo 4.

[Krenske]

Hi Mark,

Sorry for not being Active I am just about drowning in work at the moment. I literally have a stack of papers 2 feet high on my desk.

I will try to comment-

Laurent - Two perfectly equal forces will lose strength at a similar rate. They will not just sit around there and take it till they reach 0 strength though. One will quit the field, possibly both. As an example - a phalanx unit with total firepower of 80 vs 4 hordes with firepower 20 each may result in similar losses to each side (Although the phalanx almost certainly has higher experience and training values thus skewing in their favour, but lets pretend.). The phalanx will lose troops at the same base rate as the opposition (although the hordes will probably receive a bonus for surrounding the phalanx) but at a higher percentage rate. That assumes that the actual manpower strengths of the units are in a 4-1 ratio.

The units involved in the combat need to take a morale check to determine if they continue in combat. The Phalanx will suffer penalties for being outnumbered and for the casualties suffered, they will probably pass their morale check and stay for the first 3-4 rounds of combat, until losses really start weighing heavily. The opposition with 4 units will not suffer as many negatives but as there is four of them, there is a greater chance of one of them failing the check . As soon as one fails and retires then the balance of the fight swings towards the phalanx, their morale check value improves due to less damage and not being as outnumbered. In the following rounds the phalanx will do more damage than the remaining hordes and it will be even more likely that more hordes will withdraw from the fight. The last side remaining wins (It may be a costly win but it is still a win.). This is the plan for a encounter battle.

Where one side is defending and has some levels of fortification then a loss of morale will not remove a unit from the fight but will instead drop their fortification level by one. Once that hits zero then they have been pushed entirely out of the fortifications and the next failure forces them out of the fight. Most fortifications until the 1700's were of the single point but high strength type, namely castles and solitary forts etc, occasionally there would be extra large castles or towns with multi walls etc and they may reach level 2. As well some areas would have numerous castles etc and so the square could be rated as having a +1 fortification level. Around the late 1700's more extensive fortifications developed that could reach higher levels. By WW1 fortifications reached the point that they could be given a value of 4 or 5. This means that the attacker could not normally hope to breakthrough and would expend troops attriting the enemy and reducing their fortification level. The defender could then respond with reserves and block the attack and rebuild the defences. (One of the effects of attacking armour may be a check to see how many fortification levels they remove with a successfull attack.)

I think the situation you describe of 4 hordes v 1 phalanx should end up with a 2-1 win rate to the hordes but if the phalanx brings just one friend with it the battle would rapidly swing its way. 4 to 1 manpower odds are hard to overcome if the forces are equally skilled etc. Mind you if the phalanx was a veteran, super trained unit lead by Alexander against average hordes then it will give more and take more and probably win a 4-1 battle.

Mark,
the morale test you quoted is actually an old one, I have revised a chunk of the design and the page is at - http://www.usq.edu.au/users/krenske/game/g/assault.htm
I think I posted this before.

The morale check (or as I have called it the Advance/stay in combat check) itself has been given a formula as follows -

Advance test = ((morale + training + experience + (Mobility / 2) + Generals morale bonus) * 5) - 2 per 1% damage taken so far in the battle.

A high morale(10), well trained(4), veteran(4), inspirationally led (4) panzer(mob=12) division rolling into russia at the begining of Barbarossa could have a very high morale test value for that turn. In fact plugging the values into the formula gets us (10+4+4+(12/2)+4)*5 = 140%. The unit will have a manpower strength of around 100 and so will need to lose 20 strength before it will even consider stopping combat.

Quite often this test will lead the sides to fight several rounds at least before finishing the battle. I would also consider the use of a willingness to take casualties factor (maybe call it ferocity) that would alter the morale check value. This means that the commissars are lined up behind the peasant or the glorious leader has ordered the forces to stand to the last etc. This could be a player settable factor at the TF level each turn.

I can't really see the higher value is bad point you were making though.

[Mark_Everson]

Hey Krenske, thanks for the detailed comments . Sorry to hear you're drowning in work... I sure hope most of that stack disappears in a freak windstorm in your office .

I was using the link I put at the top of this thread to your model. I guess I never updated that one... will now! So anyway I was checking out the model at /game/civ/ when I should have been looking at /game/g/
I understand the current version better and it seems to make sense, thought I haven't checked it out in detail. BTW the link to the scouting model on the new page is broken. If you get a chance to fix it can you let me know here? In the meantime I've updated the link at the top of the thread so it avoids the deadlink... but its kinda a kludge.

Good Luck getting out from under the paperwork!

Mark

[This message has been edited by Mark_Everson (edited March 28, 2001).]

[Richard Bruns]

So is Fiera drawing emements, units, or Task Forces?

Would a Legion be a unit or a whole Task Force?

[Mark_Everson]

Krenske:

The part of the old (/game/civ/) page I was having trouble with (still do!) is the "Morale check at conclusion of combat cycle" section. Big numbers are good there, just like in the advance case you cited a few posts above. Then there's the Table. It has things like:

Test value__In combat
<20%________Gain 1 morale

So if my number is less than 20% I Gain Morale??? That's what I don't get. All the mods in the formula make big numbers the good ones, and then the table says small numbers are good and big ones lousy... Can you enlighten me?

[Mark_Everson]

He is drawing TFs, but we are using the most representative unit (by power) in the TF to represent it. So a TF composed of 2 units of chariots, and 2 hordes will be represented by the chariots because overall that's where the power is. However in the TF Box from demo 4, or whatever it evolves into we will show all four unit icons.

A Legion is a unit.