Military Model VI

[Mark_Everson]

Hi jbl18, I just wanted to welcome you to the forums! I am about to leave on vacation, so unfortunately I won't have time to reply to your comments for some time.

Cya,

Mark

[LDiCesare]

The build proposal sounds good. I may have some trouble implementing it, because if attacker and defenders entrench themselves, that causes a few problems, but I can probably find some way around that.

Flankereffectivity must be at least 1. What it does is divide the fronting, so:
Suppose several men with flankereffectivity N attack 1 man. 1 man is in front. N others will flank. So flanking = 1 means 2:1.
There is no plan for formations. I should already order the attackers somehowe so that the most efficient flankers flank, but that is not done yet.

jbl18,

a) Sightings should be 2 squares away

Maybe. But keep in mind the scale factor.

b) Combat engagement can occur when any unit is adjacent any other unit, rather than once they enter the same square.

I think the way to do it would be to allow units to move to the adjacent square automatically, allowing better interception...

c) A combat “half-turn” should occur as the battle is resolving; the army may retreat at any stage while giving a free shot to the opposition.

Actually there are 10 ticks per turn, and defeated units can start fleeing at each and every tick. The other side will keep fighting until they manage to get out of the square, which may take some time.

Plans include having merchants for supply lines. That's not yet coded. As for the rest (danger in terrain, etc.), this may be interesting (I know terrain damaging units is cool for gameplay). I won't comment much on it because I don't have time to code anything of that kind in the foreseeable future.

[LDiCesare]

Cross-posting form Demo 7 comments thread:

(combat) has to have a degree of uncertainty or there's no tension, and no tension means no fun.

I wonder whether I should multiply all the figures at the start of the fight by a random number between say 0.5 and 1.5 for each side to make things more random? Currently, random rolls are lost by the statistical effect of there being so many of them.
What do you think? I think that's not the only time this concern has been voiced.

[Mark_Everson]

I am going to start by reproducing Heffalump's comments that are relevant here, just so they're available in this thread.

Originally posted by Heffalump
Military model
- I liked the simultanous movement. Sure it was frustrating to 'miss' the enemy at times but it added a bit of unpredictability. Such confusion often existed in eras (unlike today) where communication as difficult and intelligence often old.
- I did not like the combat itself. Whoever has the most guys wins ... and wins in a big way from the looks of things. Aside from being patently untrue many times thoughout history, it's also boring. Combat has to have a degree of uncertainty or there's no tension, and no tension means no fun.
- How do I know what units have what particular strengths/weaknesses? I was just building things at random.
(snip)
-There seemed to be a lot of interesting things going on behind the scenes when I read the combat log. Am looking forward to it getting fleshed out.

On the issue of inadequate randomness, I agree with Heffalump's comments. Some overall randomness should be coming from the scouting and other pre-battle phases. Laurent, do those end up making much difference, or are the also lost in all the die rolls? I agree with a solution of the sort Laurent proposed, that would arbitrarily move the effective odds either way by something like a factor of 2. Maybe done using a normal distribution or some such would be best. But I don't want to lose the importance of the scouting phase etc in an overpowering effect of one random number. I think this can be fixed fairly easily after playing around with a few solutions.

Laurent, where is the most accessable documentation on unit strengths and weaknesses. Is it the military XML code, or do we have something better? There is certainly a lot of discussion of this stuff in the mil threads (and maybe some in tech), but other than his skimming them I don't know of a way for him to get that info.

[Gary Thomas]

I have (and actually have always had) a big problem with one aspect of the military model.

The actual model concerns itself with combat at a very detailed level, and does a considerable amount of calculation. However, a "military move" is supposed to be a month. I have (a while ago) played some Clash games in which the SAME battle lasted for a number of moves - that is, several months. Historically, I do not know of any non-seige battle that ever lasted more than a few days, and even the seige ones were occasional attacks. So, in a month, the battle should be well and truly over.

Also, I do not think that "retreat from battle" should be into a different square - a force that routs back 100km will always be destroyed in real life.

Personally I would like the square occupation rules to be rationalized, with an option of not fighting if neither side wants to fight.

Routed troops should remain in the same square, but unable to fight.

An army is beaten when all its troops are routed, though there should be provision for a fighting withdrawal. This would leave at least some of the losing troops in a combat worthy position if attacked next turn.

It is my view that the model contibutes nothing to the game (as a game), but could eventually be useful when the game reaches the stage of having tactical visuals. Watching the Peloponesian right wing (with the Spartiates) crumble because Epaminondas had heavily reinforced the Theban left wing would be, to say the least, entertaining.

However, the present system makes it impossible to assess the relative strengthes of the forces, and makes it extremely difficult to make sensible military decisions. The tendency is to just hurl what you have into the fray and hope.

As noted in the previous posts, God appears to be on the side of the big battalions.

There is, in fact, esentially no generalship available.

Cheers

[LDiCesare]

I have implemented something which allows the taker of the capital of a province to take the whole province. This should trigger a popup message explaining what happened.

Now to Gary's remarks (that I should have answered a long time ago).
The model does indeed a lot of calculation.
Retreat from battle includes a move order, so it is not immediate. The assumption is that the winner would keep fighting and so the loser has to flee or keep being beaten. It is perfectly possible for 2 units in the same square not to fight, although currently the decision is solely taken based on diplomatic statuses.

I don't really get what a routed unable to fight unit would mean. Would it be able to fight the next turn?

A simplification of the combat model could be useful in order to lower the amount of calculation and make it easier to understand what happens. I agree that the big amount of code I've written to implement the model provides little added value wrt a CtP2-like model.
Currently, the factors that affect fight are: attack*defense*helath, distance strength. Archers do have an effect on the fights. Then walls and siege work quite fine IMO.
Some fights last forever because the defending unit remains in the square and can only be attacked, roughly at 2 vs. 1, thus any bigger advantage (except for that provided by archers) is lost.

The scouting part is almost unused because there are very little scouts and skirmishers, and usually as many is both armies so they cancel one another.

[LDiCesare]

About taking capital implies taking the province:
In a mail, Mark said:

I would also add a threshold such that if the militia strength of the remaining squares is large compared to the invading army strength that there would be no switch. This will prevent a single anemic unit from taking a large and prosperous province all by itself.

I think it would also be possible to systematically check militia in every square so they could launch a fight by themselves, or maybe not every squares but only those which have one or several of these: capital status, walls and/or a city.

[LDiCesare]

As I said, I implemented the fact that, if you take a province capital, you lose the whole province. Units on squares in the province immediately conquer them back. This may cause trouble when there are allies/neutrals since your squares may become allied squares, but well. Also, right now, walls remain where they were and with their previous owners, which will lead to weird things. I may have to devise a new systems for walls so they are civ-less, but then their stats would be problematic.
The resulting popup's main problem is that when you take one square back from a lost province, and then lose it, the panel will show. That can be annoying. Gameplay will tell.

A way to see immediately that square X is the province capital would be welcome.

It should also be a good thing to have the player pay for moving a capital and, when a capital is conquered, there should be automatically the option for the invader to make this old capital the new capital (i.e. instead of the first square Hannibal conquered be the capital of Gaul, let it be Marsaglia).

[Mark_Everson]

Originally posted by LDiCesare
A way to see immediately that square X is the province capital would be welcome.

It should also be a good thing to have the player pay for moving a capital and, when a capital is conquered, there should be automatically the option for the invader to make this old capital the new capital (i.e. instead of the first square Hannibal conquered be the capital of Gaul, let it be Marsaglia).

I agree that both of these are good things to do.

Perhaps a small star in the civ color (maybe on a white background so it stands out) could show provincial capitals, with a larger star on the civ capital?

For the second one, we could make it such that the first city conquered in a province would automatically become the capital?

[Martin the Dane]

I have a few questions that needs to be adressed in order to make that proposed capital system work.

What about recapturing capitals?

Would a player recapturing a province automatically have his/her (provincial) capital moved to the first city he/she recaptures?
If so will this first city be a temporary or permanent provicial capital?
What hapens when the player recaptures his/her old (provicial) capital?
What hapens when the player captures the new rulers provicial capital, if this is not he same as his/her old one?

What about cities and surrounding land, when a player looses/gains control of a province by taking the provicial capital? (rememer ther is one capital for each player holding parts of a provice)

What about terrain ocupied by alied troops during a capital takeover?

To take an example:
A and B are allied and at war with C and D who are also alies. (do we have aliances implemented?)
A has a provice with a number of cities where C has captured the capital, D currrently has troops in 4 squares including one city, A and B both have troops in several squares including at leas one city each. Now A takes back his old capital. What would happend?

[wombat42]

Originally posted by Gary Thomas

It is my view that the model contibutes nothing to the game (as a game), but could eventually be useful when the game reaches the stage of having tactical visuals.
...
The tendency is to just hurl what you have into the fray and hope.

As noted in the previous posts, God appears to be on the side of the big battalions.

There is, in fact, esentially no generalship available.

Just out of curiousity... Is a smaller-scale tactical control mechanism planned for somewhere in the future of the development of Clash? Just about every other part of the game is micro-manageable right now except for military operations. Maybe a subroutine that would open a new window in which time and space were scaled down to enact individual battles?...

I would just like to suggest it, if only as a goal for future versions. It might help solve problems with the tactical realism of the game. Like I said before, everything else is pretty micro-manageable (or soon will be) except this, as far as I know - although if there is something in the works I didn't know about, I wouldn't be surprised. And it could be set as an option, with another option to revert to the standard calculations and forego the battle... Maybe you could make different levels of tactical management, from its current state through campaigns or invasions of provinces or individual battles, depending on the preference of the user.

[Mark_Everson]

Originally posted by wombat42

Just out of curiousity... Is a smaller-scale tactical control mechanism planned for somewhere in the future of the development of Clash? Just about every other part of the game is micro-manageable right now except for military operations. Maybe a subroutine that would open a new window in which time and space were scaled down to enact individual battles?...

Hi wombat, thanks for your input. One of the reasons we went with combat as it is was indeed to have the option of a tactical combat mini-game. Another reason was to enable semi-realistic descriptions of the combat and why a player won or lost. That would be a game texture and feel issue independent of a tactical combat mini-game. Given our resource constraints, though, both things are far in the future. . . Know any java programmers interested in complex games?

[wombat42]

Hmmm... Well, if I could program Java, I would be jumping to help out, but as it is...
If I do learn sometime in the near future, I'll offer my services, though.

[LDiCesare]

Tactical combat is not going to be in anytime soon unless we find someone who wants to code it (would be welcome).
I find the system quite complex and some data in the model are not very precise (mobility in particular). Sometimes I wish I had coded something simpler.

[wombat42]

By the way - is there any way to look at the source code for the game? I'm having mucho trouble trying to open stuff in Notepad...

[Mark_Everson]

Hi Wombat:

Send me an email saying you want the source code and I'll reply. Best to have an acct that doesn't freak for attachments somewhat larger than 1MB.

[yellowdaddy]

Gary said:

Also, I do not think that "retreat from battle" should be into a different square - a force that routs back 100km will always be destroyed in real life.

Personally I would like the square occupation rules to be rationalized, with an option of not fighting if neither side wants to fight.

Routed troops should remain in the same square, but unable to fight.

I take it that with your "non-square" Garymap, this will no longer apply? (at least not precisely the same way)

[Gary Thomas]

Under the polygon system every unit will have co-ordinates, not merely a square occupancy. So a unit could rout back 5km, for example. If that put them in another polygon, they would then be in the new polygon.

Cheers

[LDiCesare]

Note I have lots of ideas about things that could be redone in the military model, in particular how morale works, but then it's working for now so that's not top priority.

[alms66]

As suggested here, by Mark, I'm bumping this thread to discuss changing the current military model.

Other useful discussions:
Fortifications & Sieges
Military Coding Status
Military Interface & AI

The first thing we need to figure out is what we all agree needs to be fixed. To that End, we should make a list of what we each believe to be wrong with the model.

At the moment, my main gripes are:
1. The model is, in general, too complicated. It can be vastly simplified and achieve the same effect (realistic combat outcomes and casualties), given that the player sees, and wants to know, but one thing, “Who won?”
2. The model focuses too much on tactical details, rather than strategic ones, which… again, brings us to the one vital question, “Who won?,” as the most important part of the combat, the second being “By how Much?” or casualties, of course.
3. Armies unable to stop oncoming units. This is a major issue with the current demo. I hate it when I see a unit coming, but on the turn when I should ‘hit’ it, it ends up one or two squares behind me. This stems from simultaneous movement (which I support and don’t want to get rid of), and scouting (which I feel is an unneeded complication in 99.9% of cases).
4. Automate non-combat units into a generic cost of running the army. Simple. Easy. Enough said.
5. Lack of ‘Superiority’ concept (Air superiority, cavalry superiority, etc.)

One and two are really big gripes, the others are smaller, but still main gripes.

Edit-
Thus, I'd be in favor of starting from scratch on the design, reusing as much code as possible to match the new design, and sucking it up as a worthwhile loss on code that can't be saved.

[alms66]

I went ahead and typed up (yes, it was hand written on paper) the model I mentioned that I had. It's very rough, as copying the thing in it's entirety would have been pointless at this point, but it should be understandable enough for people to say what they think about the basics of it. It's in "rtf" so everyone should be able to open it, let me know if you have trouble with it.

[Mark_Everson]

I checked it out Alms, but don't have much intelligent to say about it at this point. I do agree that there should be bonuses or other benefits for cavalry, etc. superiority as you have in your writeup and comments above.

As I've said before, I'm not in favor of ditching the whole system. But I think there are certainly some fixes that can be applied to make it much better. I haven't got my list of ToDos for the near future for the mil model together yet. Two big things:
1. As you said. . . auto ways to chase down enemy military units in a civs terrain. Also a better handling of the status of territory that is taken over and then the enemy leaves. Many times the square should revert to the previous civ's control.
2. A better handling of supply. I think units not in good supply should move much slower, since the have to forage for food. They should also cause economic damage to the areas through which they travel.

For the record, I just wanted to post My first Clash Military Model because it also has some good ideas (IMHO ) that we could try to include in the existing model. I have not put up the whole old clash web site, so the links at the bottom of that page don't work.

[edit] much added to the middle of the post

[alms66]

Originally posted by Mark_Everson
I do agree that there should be bonuses or other benefits for cavalry, etc. superiority as you have in your writeup and comments above.

This is certainly a missing component in the current model.

Originally posted by Mark_Everson
1. As you said. . . auto ways to chase down enemy military units in a civs terrain. Also a better handling of the status of territory that is taken over and then the enemy leaves. Many times the square should revert to the previous civ's control.

I’d even go so far as to set up a provincial defense force, which intercepts enemy units as they enter the province. This would cut down on micromanagement of armies in a BIG way. The player shouldn't have to do anything to set this up, it should be automatic for all units stationed in an owned province.

Originally posted by Mark_Everson
2. A better handling of supply. I think units not in good supply should move much slower, since the have to forage for food. They should also cause economic damage to the areas through which they travel.

As a quick fix, I’d implement a cost factor for being away from an owned tile. For example, if a unit has 2 movement points and it costs 1 to cross flat tiles, he could go on forever, however, with a cost factor of [.25] it would cost 1.25 movement to cross the first flat tile outside of owned territory, 1.5 for the second, 1.75 for the third, etc., until the cost rose to a point where the unit could no longer move (effectively). There’d also have to be an attrition value which rose in the same manor, killing off or forcing settlement of men in the army.

Originally posted by Mark_Everson
For the record, I just wanted to post My first Clash Military Model because it also has some good ideas (IMHO ) that we could try to include in the existing model.

If the player was able to say “I need 20% of my forces to attack Spain, and 35% to guard my homeland, another 35% to guard my colonies, and 10% to secure my Spanish conquests” by using some sort of football-style interface, as you propose, this would be a great tool for players.

[LDiCesare]

Let me stress quickly the points I consider bad in the current model/code:
- Not random enough
- Not enough importance given to morale
- Cavalry has little effect wrt infantry. The only difference that is valid at this point is archers/non archers (there's also artillery for siege which I think is OK).
- Several variables are unused or not very useful, and could be either ditched or reworked
About the point:

2. The model focuses too much on tactical details, rather than strategic ones, which… again, brings us to the one vital question, “Who won?,” as the most important part of the combat, the second being “By how Much?” or casualties, of course.

I slightly disagree. We could show some tactical results so the player understands what happened (showing the battle the way CtP does it, or Dominions, for instance, or giving some details on what went wrong).

Armies unable to stop oncoming units.

This probably has more to do with a set of bugs than a revision of the model itself. The only thing I can think of which could be useful is an order to 'block/intercept' instead of move. That isn't part of the military model per se, but the movement code.
I'm not sure I get your point about scouting. Again, I think it's not entirely military. Spying should give info on what forces are in far away squares, but we haven't implemented spying yet.

Automate non-combat units into a generic cost of running the army. Simple. Easy. Enough said.

I don't get this one either since the non combat elements are not shown currently, they are just abstracted into a cost of the elements.

[alms66]

Originally posted by LDiCesare
I'm not sure I get your point about scouting. Again, I think it's not entirely military. Spying should give info on what forces are in far away squares, but we haven't implemented spying yet.

I don't get this one either since the non combat elements are not shown currently, they are just abstracted into a cost of the elements.

Scouting as a phase in combat is useless. If two units enter a square, they should fight (if enemies), not scout to see if they notice each other. Spying, I agree should provide benefits as you say, but it has nothing to do with the combat itself.

Non-combat elements are added to various units. Get rid of them, and replace them with a cash cost to maintain the unit. This way, players don't have to worry about engineers, miners, etc., but just swordsmen, spearmen and siege towers, etc. I know that currently the player doesn't even see them, but a scenario designer does. Personally, I'd rather drop them all together, or simply increase the cost of all units by a tiny bit to account for the 'extras' in the military. It's not that major an issue with me, and I can live with it, I suppose, if that's what people want.

[Mark_Everson]

Just for the record, I like the scouting phase, and also the ability to add the support elements to units.

[LDiCesare]

Engineers are not non combat. They are just useless except in sieges, and will fight if noone else is left.
Scouting phase adds details that aren't seen by the player and doesn't affect the result of the battle a lot, so I agree it could be ditched or reworked. It changes about 5% of the strength of your army, then manoeuvering about 15%, but these figures are far too small to have much effect considering that, in practice, the armies are usually about matched in these areas.

[Mark_Everson]

Hi Laurent:

Originally posted by LDiCesare
Let me stress quickly the points I consider bad in the current model/code:
- Not random enough
(snip)

From my recent play testing I think we need to increase the randomness further for demo eight. It is still the case that when you have a 2:1 advantage you are guaranteed to win. Can you just double the easily-modifiable random factors in the code so that we can evaluate the results? if I recall correctly the randomness in the Scouting say is should be easy to change. If that's not practical, anything you can do to increase the randomness somewhat would be valuable.

[LDiCesare]

It's pretty easy to add some randomness just about everywhere. I'll do that. I think the current state is already better than what it used to be, as I managed to be surprised by the outcome of a battle when I played this week, which wasn't usually the case.

[Mark_Everson]

This is a crosspost from the Fixing the enemy units behind-the-lines problem thread.

After my most recent set of play testing last weekend I had the following thoughts:

1. Eliminate all units that fall to below 10% of starting strength. (If a similar unit exists in that square that could take the reinforcements that it should get them.)

2. Any unit that is in a fight at worse than 10:1 odds should be annihilated. (There may be some exceptions, for example if the weaker force has higher mobility it should have some chance of getting away.)

Number 2 above points out what I think is a big flaw in the combat system. It restricts the number of units that can fight a given unit to a very small number. (I think it is either one or two.) In the real world there are many ways were the excess troops can make a huge amount of difference. I think that at a minimum the combat matching should allow greater odds. Perhaps such a change is beyond the scope for demo 8, so I will just copy this part of my post over to the military thread.