Military Model V

[Mark_Everson]

Gary:

We can certainly try most of this stuff out. But I think its all longer-term. If there are any restrictions in the code to making this happen, then we should try to get rid of them. But as Laurent said, much of it is doable, much with no or very small tweaks.

Originally posted by Gary Thomas
The other is that the present model precludes a player making some of the significant changes that have affected military history. The Marian reforms of the Roman army, for example, would have to be presented externally under the present model ("OK, now you are able to build a manipular legion"). I like the idea of players being able to experiment with different organizational patterns (or not, if they don't wan to).

I'm not sure exactly how this would work. We had some extensive discussions a long time ago, I think in the previous mil thread, about military doctrines for elements, units, or civs. And you could express the Marian reforms as a change to a doctrine representing more flexibility. But you're the mil lead, go ahead and throw together a spec on how this would work in your copious spare time. Gotcha You might want to skim the old threads first, since they do contain a lot of good ideas. But again I think this is a thing for four or five months from now...

Originally posted by LDiCesare
Training:
I will build units with 0 training. It is also easy to put experience in the code at the same level as training except gained differently, (even though I don't like it much).

I still think building all units with 0 training would be very bad. The default level of training in building a unit should be set similarly to the way it is now. A Roman Legion or tank battalion with no training is Militarily Useless. Several experiments of this type have been carried out over history, with the predictable results. I do not think Clash will be improved in any way if we give the player the flexibility to build useless units! OTOH units like a horde can be built with very little training, and should be reasonably successful at fighting that way.

If we do the slider for training idea, we can let the slider go all the way to 0, and the player can experience no-training for themselves. This ties in with the idea of having a function that determines the result of training. Types of units with complicated or flexible structure should be Severly penalized at the low-training end.

[colorrr]

Mark, you are right it was a language issue, recruits are the input, soldiers are the ouput. I was referring to the output.

I think having two scales for training and experience would be a little confusing.
I like better having a 0-100 point system for experience, where units start out at f.instance 60 and can be trained to a max of 80 -/+ maybe depending upon tech/cultural attributes. To reach 100 experience the unit would have to see real combat.
A higher level of training could also be achieved by introducing live fire exercising, like they did with Waffen SS troops during the war.
With live fire I do not mean firing with live bullets at fixed paper targets, but doing training in various situations while real bullets are fired above and around their heads.
Some would die during training but at the advantage of the men already being baptised by fire, and less likely to loose cohesion the first time they see real combat.

When a unit looses men in combat and need replacements, these replacements could come in at the 60 exp leve, raising the total men in the unit, but lowering the experience (and morale?).

Training after a unit is formed is often executed in the field by commanding officers of the unit itself, so any training facility would not really be necessary to do it. I think units should be able to train, or automatically gain experience each turn, without the presence of any military training facilities, but the rate should be low. The presence of training facilities in the civ could accelerate that rate.

I agree with Gary that the player should be able to control the default amount of men that goes into a unit (it's TOE).
I don't think it should be by a slider. I propose something similar to todays structure.

If I am not wrong, the SNLF units were battalion or regimental sized. So even if they were not used in division size numbers, they are still represented by a military structure that is a building block of a division.

In the game we could allow the smallest unit to be a battalion. The different arms always come in battalion formations that are purely of their own arm. Then we allow the user to customize the composition of the bigger formations by using these battalions as building blocks. F. instance, The Phalanx unit could be 9 battalions of spearmen, 2 battalions of skirmishers.

Whenever a player wishes to build his military he trains x number of battalions of that type.

Maybe we could leave out completely the idea of getting a new complete military unit with technology.
New techs would instead give a new military building blocks: spearmen, riders, musketmen, riflemen, engineers, armoured reconnaisance etc.
The player would customize his own units using these building blocks.

Maybe too much micromanagement?
That could be resolved with new techs also bringing new predefined structures that the player could use, modify or throw away.

[LDiCesare]

Training + Experience is currently modeled (and half-coded) as 0-5 + 0-5, so you have a 0-10 scale for the sum of both, and cannot get more than 5/10 without seeing real fights.
I think real fire exercises are a bit too detailed to include or model.

Training is considered training beyond minimal value. That means when building a unit with 0 training, you do have some services cost. You don't get bowmen by just giving them arrows and bows and saying go ahead, shoot. You at least drill them on how to string the bow, aim and shoot. That is 0 training. More aiming/shooting will raise to 1-2, and synchronisation, explaining how to shoot at what distance will increase to 3-4. Intensive drilling will lead to 5. So level 0 still requires more training in the sense of services than a warrior (anyone can use a club). Thus the training I am interested in is not the economic cost of training, but the military value of training beyond base drilling. Minimum required training is reflected only in the cost of the unit, as the services cost. Indeed, archers or spears need more services than warriors, and mounted troops need training in order not to fall from their mount too, so they have higher training cost than their foot equivalent.

[Mark_Everson]

Laurent, sorry I misunderstood you on the minimum training level. I do think that having five levels in the code is too restrictive. We must Avoid those big discontinuous jumps IMO. It should be a float like everything else, with a function determining the effectiveness. In terms of reporting to the player, five levels is just fine!

Colorrr, I agree with you that training in the field is important also. In fact one of the big determining things between army qualities IMO was that the good armies would train continuously whereas the worse ones would just slack off and do fun things when there wasn't a battle. Another possible place for a social model connection? And as you say training in a dedicated facility would of course be more productive.

[Gary Thomas]

In my comments (somewhere or other, probably in emails when the forum was defunct) I said that the instruction from the economics sector to build a unit should not know about training.

Unfortunately I have lost the text of my email on the subject (perhaps Mark or Laurent can dig it out). Generally my perceptions was something like this:

1. Administration constructs an order to build a unit (a "buildable" object).

2. This order is passed to the economic sector, which queues it.

3. Eventually (or immediately) the order is executed by passing the order to the (newly created) buildable object.

4. The buildable object (the new unit) informs the originator of the order that it has been built (or not, if building is not possible), the administration then informs the economics sector that the order has been completed.

5. The buildable object knows how to deploy itself.

The part of this that is missing is that the original order could be "build and train a legion", with the training level specified at that point.

Cheers

[LDiCesare]

Mark, training is a float, it is just taht it values are between 0 and 5, but it can be 3.42 if you like. (I bet only the AI would do that, unless some player fancies that 3.141592 is a pretty value for training.) The figures/examples I gave were just interpretation in English of the figure.

Gary, I think I still have your mail somewhere, do you want me to post it here?
I do agree with you anyway. Econ mustn't know training, but something between the UI and the econ must know it, and can either give correct input to the econ, so the buildable result will have the training needed, or create a unit and, when it is created, issue orders for drilling, which are econ orders too.

[LDiCesare]

Fleeing

The military model models flight (fleeing). However, I have trouble presenting this as an output of the model to the player.
The reason is that there can be up to 10 fights in one square in a turn (10 ticks), thus a unit may stand its ground 5 times, then flee three times, then stay again.
Currently, a fleeing element is out of the fight, but will remain in the square, ready to fight during next tick or turn.
Thus it is very difficult to tell the player what happened, as elements that fled may have been killed later and some fled several times but not always. Thus I feel I can't tell the player in the CombatReports in any understandable way.

However, the player can see units fled by seeing them move away:
I tried to have fleeing units move back to their previous square. That had 2 bad consequences:
1) It screws up the TFs, leading to drawing problems. That can be fixed, but: I am unsure the player would like to have his 10-unit TF split in one 5-unit remaining and 5 1-unit TFs in the neighboring square because each unit fled at a different tick.
So maybe there is something to see between the TF box and the command system so that the drawings follow the command system. Trouble would also appear displaying a scattered Command's path.
2) This may be subject to balance. I tested with one element fleeing means the unit flees. We can decide only if 50% of elements flee does the unit flee. Anyway, fleeing is pretty common, particularly when outnumbered. Thus, armies tend to remain on the screen much longer, because they flee when they see a huge horde. To eliminate totally enemy units, you must use a small part of your army that looks not too dangerous so the others will be lured into fighting and killed, or get archers who always get a chance to shoot at opponents before someone can flee. It means mopping up forces to chase small armies. Note it can make guerilla worthwhile, and forces you to split your huge TFs so it is not necessarily bad.

Based on this input, do you think I should move the fleeing units back to their previous square? I think I should, even considering the whole bunch of problems I just described, because it is the only way to have a blocus. Currently, if you put an army in a square, opponents can go thru, sustaining some petty damage, but go thru nonetheless.
Also, they will be able to move back to a previous square in one tick, which is pretty fast.
Another option would be that we could stage flight in 3 ways:
Suppose a unit has 100%elements:
If less than 30% flee, it stands its ground and remains in square next turn.
If less than 80% flee, it moves to previous square immediately.
If more than 80% flee, it surrenders and just disappears. This could be based on another morale check, though, but battles have been gained without fighting, and armies disbanded, so it seems correct, and would avoid chasing one-element damaged units round the countryside.

What do you think?

[Mark_Everson]

I think the solution at the bottom of your post is a good workable one for the near future. Eventually there should be a commander AI that decides after a battle whether it wants to stick around or not. We will have to work on that sometime soon. But until then I like your solution! We definitely don't want Clash looking like a fight-to-the-death system!

[Mark_Everson]

Two military orders that were in demo 4 that I'd like to see again… I may have mentioned them before, but don't think so.

Support (this was actually a default behavior rather than an order)
Commands should support friendly commands that are defending in adjacent square combats. After a tick, when it is determined where combats will happen, any friendly units that are in adjacent squares can attempt to join the combat. To do this they must have not moved the previous tick. This is to simulate detection of the enemy army nearing before the actual combat, and the massing of any nearby forces available before combat. Eventually the model should be much more detailed than this. In the old demo TFs attempting to support friendly units in combat stood like a 40% chance of being able to make it to the battle in time. If we get agreement between all the military 'stakeholders' we can work out a spec

Sentry TFs try to support adjacent friendly units or squares if they come under attack. Having a unit on Sentry duty should guarantee it can support a friendly force in combat. After that combat the sentry TFs should automatically return to the original square on which they were stationed. This order allows for construction of decent pickets. These are most useful against barbarian raids, and other light battles, which we don't have much of yet. But I'd like to see them sometime.

[Gary Thomas]

I don't really like the idea of 5000 men moving 100 km because some troops there are under attack. The actual logistics of the move would mean that it was all over by the time they got there. In the meantime, some baddies could have sneaked through where they were supposed to be guarding.

If these units are to be regarded like Civ2 units - sort of chess pieces, there isn't much problem. If we want to have it a bit closer to reality, we have to realize that a square is quite a large distance for 5000 men to move. Another factor to consider is that a substantial majority of historical battles cannot be represented in the Clash system, because the numbers involved were too small. In the Dark Ages in particular, 5000 was quite a large army, getting close to the limits of most areas to support logistically.

Cheers

[Mark_Everson]

Hey Gary:

Originally posted by Gary Thomas
I don't really like the idea of 5000 men moving 100 km because some troops there are under attack.

"This is to simulate detection of the enemy army nearing before the actual combat, and the massing of any nearby forces available before combat." Which of these words didn't you understand? This does Not begin when the fight starts, waiting for the fight to start is what we do in the code so it doesn't require AI that isn't there yet. The mechanism is simulating when side A knows side B is coming and is still some ways off. The estimation of When different commands find out about an impending battle, and how far they could move could be quite sophisticated, but not at this point. The AI could make assessments given current info and automatically issue orders for nearby troops to mass as appropriate previous to a battle. But we're not there yet.

If these units are to be regarded like Civ2 units - sort of chess pieces, there isn't much problem.

We can see, but this may be one of those cases where even if it weren't realistic (and I do not concede that point) it would be worth doing for gameplay purposes. Essentially it is a technique to reduce micromanagement.

If we want to have it a bit closer to reality, we have to realize that a square is quite a large distance for 5000 men to move. Another factor to consider is that a substantial majority of historical battles cannot be represented in the Clash system, because the numbers involved were too small. In the Dark Ages in particular, 5000 was quite a large army, getting close to the limits of most areas to support logistically.

We already dropped the 5000-men assumption IIRC. TFs can have any number of elements, composed of any number of men is the definition I thought we adopted for long-term use.

[Lord God Jinnai]

For nomads: thge non-miliart units cannot attack, but should have a defensive value.

Militia created from nomads and minor civs/despotic civs/narbarians need not rely on supply routes and shouldn't get as much attrition when living oiff the land as they are more used to it...mercenary iunits hired from here should be at your discretion (your meaning who's in charge of this model) as to wheter they should apply with those rules or not.

[LDiCesare]

Fleeing

I coded the 30%/80% flight stuff, but it doesn't work. Here is why:
Units flee all the time, and that is how fights end. In one tick, the fight ends when one of three conditions is met:
-One side is dead
-One side has fled
-A certain number of rounds has passed.
The problem is currently, rounds end because elements flee. Units are destroyed only because there are ten ticks in a turn, and after several ticks fighting, they get killed.
Tuning the 80% margin changes nothing, as 100% of remaining units flee.
I added another control, checking morale afterwards, based on the number of surviving elements in the unit, but that still gives outrageous results, like all units surrendering on the second tick of fight instead of the first.
I'd like to be able to show that fleeing exists, and possibly give it more effect than it currently has, but I am running out of ideas as to how to deal with it.
Any suggestions? I will try tweaking numbers around, but anything more clever would be better.

[Mark_Everson]

Re: Fleeing

Hi Laurent:

I will try to make some suggestions that I hope help. However I don't understand the line below.

Originally posted by LDiCesare
Tuning the 80% margin changes nothing, as 100% of remaining units flee.

There are some other things I don't completely get, but here goes...

If I understand, I think there needs to be a high-level decision to tactically withdraw made before each tick's fighting. If a side is almost certain to lose it should attempt to withdraw. Note this is not the same as units fleeing, but I don't think you can show that to the player, other than as a combat statistic for the last battle in a square. This is an attempt for the whole army involved (on one side) to disengage, so you don't have to worry about splitting TFs. I expect if the odds are more than 2:1 against it a side should want to disengage. They're Not going to accomplish anything anyway. Of course eventually there will be orders that prevent attempts to disengage and other details.

When a side attempts to leave the square, it can for now, get a 50% or so chance. We can get more detailed later, with things like "horsemen can always get away from infantry" etc. If they Make the chance to leave, then they immediately get a temporary protected status that prevents combat, and begin moving to another square. If they were the attacker, I think for now they should have to return to the square they came from. I guess for now a defender can go anywhere they want (pick random if no friendly forces adjacent).

I think this will handle the issues you present, but of course only if I understand them!

You mentioned before that even strong forces don't "block" enemies from moving through them. I'm not sure how this happens since the sides should fight for at least 2-3 ticks. But assuming that isn't enough, I propose a new rule. When you try to move through a square where there is opposition, you can only continue your movement if you win the fight. Any time before victory you can go back to the square you came from, if you disengage as discussed above. But the only time you can continue moving forward is if the overall battle is won by you.

Hope this helps! If you like the rule changes, we'll need to see what Gary thinks too.

[LDiCesare]

I put the oddsToAttack/oddsToDefend stuff back in use, so it allows modelling the flights. These values are described in the "tactical" order given the unit.
For now, orders are not given to units, so their default order is their preferred order, which is ranged fight if available, else sentry. Sentry means you don't attack, so that is poor. I can either change all values so that for the moment, fights happen at strength ratios between 0.5 and 2, or we can plug in orders for units now. The first one requires much less code, of course. It can be interesting, though, as it would allow fortify orders, and giving various levels of attack ratios (as in D4). What do you think?