Military Model V

[LDiCesare]

I recently ran a batch test to check efficiency of support units (read archers).
Here are the results:
I ran 3 tests, a tournament between bowmen, hordes and phalanxes. As phalanxes are stronger, there were 10 bowmen, 10 hordes, 7 phalanxes.
The results were consistently:
bowmen beat phalanxes beat hordes beat archers.
I like it because it means archers outnumbering phalanxes can pound them to dirt from afar (effective support fire) but they are smashed by hordes if these can get close, even though hordes are less powerful than phalanxes in close combat.

Now does anybody have ideas about how to give reasonable prices (econ) to these units so that there be no one uber-unit?
In particular:
If unit U1 beats U2 unless they are outnumbered X vs 1, should they cost X times U2 or less? (X times means you'd get units faster building U2).
How much do higher tech units cost?
They usually have an advantage in terms of fight outcome, but should they be cheaper? If one knight is worth 2 light cavalries in combat and costs twice as much but additionnally requires extra tech, you'd better research other techs and build more light cavalries?
Decreasing the population when building a new unit offsets some of these concerns a bit, but I'd be for having smaller prices for stronger/higher tech units because -you wait more before you get them -it makes it reasonable to research new military techs.

[Mark_Everson]

Hi Laurent:

That was an good experiment to do! First I'll answer your question on face value, not considering reality. IMO if the resuts for 10A/ 10H/ 7P come down to rock-paper-scissors, then the cost to create all these armies should be equal! Since they give equal expectations of a win game balance demands it... giving costs using an arbitrary scale of 1:1:1.4 respectively per unit or so. Now of course this analysis ignores a lot of factors, like how effective are these units as part of a combined force. In other words given an assumed combined force of 3/3/2 what is the incremental benefit of adding one unit of each type? This may alter the cost. Also since we have multiple basic goods going into a unit we'd expect the Archers to have proportionally more Services cost (lots of practice) and the horde less, etc. This means that for different players with different economies it may be more advantageous to build one than the others considering the cost of those basic goods in the civ.

Now Reality (or our pale version of it)
We like game balance, and we as a project have a predisposition to reality (if the reality isn't too expensive, in terms of not killing fun...). So my first question would be, are these the results you'd expect for similar armies in classical antiquity? I don't care if things are a little off reality, but I'd prefer not to have people who know something about ancient battles laughing. So it would be good to get someone who knows their stuff look at it, or maybe simulate a few ancient battles? Another point is that your experiment is, I bet, assuming good flat ground. I expect the results would be completely different in woods, where the horde would win hands-down all battles. So in future tests you may want to have a suite of 3 or 4 terrain types to use.

After all my comments about refinements etc. I think your test fight was a really good idea. It will allow us a good initial way to balance costs of units. In terms of practicality, I think we should just assign the costs to make units roughly cost-effective in terms of battle outcomes. Then we can, for fine-tuning get feedback from playtesters that "I always build X" know that X should be made a bit more expensive.

Great Job

[LDiCesare]

Another point is that your experiment is, I bet, assuming good flat ground. I expect the results would be completely different in woods, where the horde would win
hands-down all battles. So in future tests you may want to have a suite of 3 or 4 terrain types to use.

Well, I still have to put in the terrain mods and then to decide on values for the various terrain types... I am not sure what had been specified for these, although I am sure that rough terrain should lower the effect of support fire (try shooting an arrow at someone in a wood). The phalanx would lose cohesion, but again, I am not sure how to handle it.

are these the results you'd expect for similar armies in classical antiquity?

As far as horde/phalanx are concerned, this is ok. Paul said 1 phalanx should be able to resist Persian warbands twice as numerous, so 10/7 is correct. The question is 10/6, 10/5. I didn't check the limit. As for archers, it is a bit hard to say, because I don't know of battles of only archers vs. something, and I am really knowledgeable only with longbowmen and crossbow armies.

We have a lot of figures to tweak here anyway. I'll try to make more tests and dress up a table of relative unit efficiency.

[Mark_Everson]

Originally posted by LDiCesare
Well, I still have to put in the terrain mods and then to decide on values for the various terrain types... I am not sure what had been specified for these, although I am sure that rough terrain should lower the effect of support fire (try shooting an arrow at someone in a wood). The phalanx would lose cohesion, but again, I am not sure how to handle it.

I didn't phrase my bit well. The results Should be very different in woods, but I'm not sure the model is there yet...

Looks like a job for Gary in the near future . I think we can get far with some fairly simple assumtions correlating unit effectiveness (attack str) vs. unit type and terrain. There was a table a bit ago Gary put up for movement, maybe we can use something similar to that.

[Gary Thomas]

How come all of your posts end with "job for Gary" or equivalent expression?

Cheers

[Mark_Everson]

Most of my posts don't even mention you explicitly. You just notice the more painful ones

[LDiCesare]

Combat stats

Here is a summary of batches I ran to evaluate the various strengths of units.
Based on the first results, I changed a few values and put costs which were approximately proportional to units effectiveness (costs being element-wise).
Units are classed in the number of WarriorBands needed to match evenly 10 units of one kind. So WarriorBand is worth 10:
Bowmen (8) WarriorBand (10) Horde (12) Chariot (14) Phalanx (22) Light Cavalry (22) Legion (23).
Legion is 7 heavy spears, 2 skirmishers and 1 engineer, which may not be accurate. I just wanted to throw an engineer in for field defense code.
Bowmen are low on the scale, but they are always less efficient than other units in melee. The figures are a bit complicated, but they translate as: You need from 2 to 5 bowmen in support of weak front units to beat 10 of anything.
I nerfed the chariot down because chariots were replaced by cavalry when people learned to mount, so I guessed they shouldn't be too powerful.
What do you think of those figures (please note there is no defense/offense bonus considered in this test, which takes place on "neutral ground")? Should a phalanx be more or less strong than a light cavalry for instance (currently, phalanx is slightly weaker but defends slightly better against archers)?
Bowmen stats: Bowmen tip a fight for a weaker front line when there are the following number in support against 10 opponents:
WarriorBand: 2, Horde: 3 or 4, chariot: 3, Light Cavalry: 4, Phalanx: 4 or 5, Legion: 5.

[Mark_Everson]

Hi Laurent:

I think your approach is great and the figures are reasonably good within my knowledge about such things. But I don't think you should make the cost so that everything is even. Remember there are substantial investments in tech needed to get to the ones on the right. If I have worked as a player to increased tech, I should get some Benefit. So I would leave the contemporaneous units like you have them, but have cost effectiveness go up as you move in clear technological steps to the right. I have no idea about light horse vs phalanx...

[Gary Thomas]

Laurent:
My email access is shaky, so I thought I would check with you through the forum.

What is the present state of ships? Can we have ship battles?

Is transport working? Can I add load/unload options to the TF popup menu? If so, what is the call I need to make?

I realize that you probably told me all this, but I don't have access to my previous emails.

Cheers

[LDiCesare]

Gary, here are the classes/methods to get on/down ships:
MovementElement.isCarrier() returns a CarrierElement. Only units that can carry other units return a non-null CarrierElement. It is then used to unload on a given square (you have to check the square is adjacent to the unit because I just teleport the armies there).
To get on board, use MovementElement.board(CarrierUnit). It will return true if it managed to board (i.e. had enough space). There are of course various methods to check there is enough space, know how much there is, list units and elements.
As for naval combat, it is coded, and sunk ships should cause the death of carried units, but that is untested. Note that it is actually the ground-fight on ships, since this only covers ancient ships.
I didn't add a roundboat so the trireme still carries units, because we don't have an image for roundboats.

You can also look at game.military.test.testBoard if you need some example.

[Mark_Everson]

D5.1 Chariots Suck

Hi all:

I complained about the chariot movement restrictions before, when Gary put up the movement capabilities table, and I'm complainin' again

Chariot moves are Waaay too restrictive. At a minimum they should have allowable but reduced movement in hills, and possibly broken. Right now Chariots are only of any good if you move over completely flat land. In Delenda the Only thing I can use chariots for is a homeland-defense rapid-reaction force, because I can't Get chariots to the enemy! Even if this were realistic (and I don't think it is) it is no Fun and should be changed.

[Richard Bruns]

Note on archers: In antiquity the power of bowmen varied greatly based on the skills of the general. For example Surenas used Parthian horse archers to annihilate a Roman legion with seven times as many men, but when fighting archers in places like Gaul the legions could butcher enormous enemy armies.

To simulate this, bowmen should probably start weak but recieve a big boost as your Military Tactics technology increases.

How do you simulate Tactics in your battles anyway? Does it do anything, or is it only a requirement to get units at present?

What's the status of generals in the character model? Can they provide a temporary boost to the Tactics technology that only affects the army they lead?

By the way, this looks good overall. I am wondering; where do spearmen and javelin throwing auxiliaries come in? Do you simulate the fact that Roman legionnares had about half a dozen javelins per person that they tossed before they entered melee combat?

[LDiCesare]

Are you sure archers skill depended on general? I would bet on equipment/training. In Gaul, celts were more proficient with slings than bows, and used short bows of poor quality. Slings are quite slow, too, and take a lot of place, and are bad for ballistics shots so that was not very efficient. Did the Paths use composite bows? A composite bow is much much more powerful than a single-curvature bow, but they weren't known by latins celts or germans (greeks had them, but didn't use them much, while persians, indians used them a lot).
Tactics are not really modelled, but there are scouting and manoeuver phases which give a bonus. Generals have some influence in the model over these phases in particular.
As for spearmen and javelins, there are 2 things in the model which allow to model it:
Skirmisher elements on one side (which are in D5), and fight at various range (which is not yet coded but I'd like to do it next): All elements have attack ratings at various ranges, and attack at that range, then go forwards if their morale didn't break, fight at closer range and so on. This is in addition to support fire like archers or artillery can provide.

[Mark_Everson]

Hey Richard:

I think it is more the mobility of horse archers that is the difference between the Parthian and earlier cases. Mongol horse archers could beat just about any contemporary force. And as Laurent says training and equipment make a big difference too. IMO all these are more important than generalship, although certainly lousy generalship can screw up just about any advantage.

Military Tactics can certainly be made to somewhat influence numerically unit performance such as Attack... I'm not sure if we have a consensus to do that yet, but its certainly an alternative. I think we do want a civ that has been using musketeers in combat for quite some time to be more effective than a civ that has just become able to use them. If Tactics is how we do that, I've got no problem with it.

Generals are going to have to wait a bit, as will any specific tactics you can issue to armies (If we ever go that way). Those factors are IMO in the second or third tier of things we need to do. But at some point in the future we can play with the sort of things you have in mind.

[Richard Bruns]

You are right about the mobility and power of the bows; but I still remember that only a few generals used the bowmen properly.

You mentioned that units fight at one range and then move closer. What if it is to their advantage not to move closer? If a unit has long-range firepower and more mobility than its opponent, it seems like it keep shooting at that range until it runs out of ammunition.