AU Mod: Experience levels

[pvzh]

We all know a problem of AI: regular troops. Human players builds barracks and have mostly vets. When Nationalism comes along, conscripts become a widespread use by AI, with little effect though.

What about to add 1 HP to every experince level, so it will be 3/4/5/6 for conscript/regular/veteran/elite instead of 2/3/4/5?

Side effects would be it will somewhat weaken millitaristic trait and will create possibility for viable use of conscripts too.

[nbarclay]

I view that as too major and fundamental a change from the stock rules.

[punkbass2000]

I agree. Plus the mil trait would just about die if it were any weaker, IMO. Especially as this is about its only advantage over Sea.

[Modo44]

The only way to keep the Militaristic trait useful would be to change the HP values even more. Adding 2 HP to the Elites and only 1 HP to other experience levels does help Militaristic civs retain their advantage. And yes, it seems very drastic concerning the AU Mod philosophy.

[Theseus]

Well, first off, pvzh, I want to again compliment you on the suggestion re the SoZ and the Knights Templar acting as barracks. I think that should help, and assuming we incorporate such change we should prolly look for some gameplay impact on AI civs before making further changes in this regard.

That said, I've been playing with HP values a bit in the AU AC Mod... I am becoming enamored of the idea of an extra HP for building-generated units and possibly for UUs as well. Something for future reference.

[lockstep]

I like the idea because it would somewhat reduce the 'luck' factor in combat, but it hurts the Militaristic trait. So ... no.

And as Theseus said, the 'veteran units' flag for the Statue of Zeus and Knight's Templar was a brillant idea.

[alexman]

Increasing the HP for all experience levels would have other side-effects. For example, since 'extreme' outcomes would be less likely, the change would weaken Horsemen relative to Swordsmen.

Also, a 3-HP unit versus a 4-HP unit is not likely to do better than a 4-HP unit versus a 5-HP unit in general. Increasing the HP of both units just helps the strongest unit, after modifiers. Since that is often the human player (the AI usually relies on unit quantity as opposed to unit quality), increasing the HP would actually hurt the AI.

[pvzh]

But with base 3 instead of base 2, unit quality has LESS impact! Barracks gives 33% increase from 3 to 4, but with a base 3 the increase will be only 25% from 4 to 5. That is the whole point of suggestion, since human player is using vets! And more importantly vets only 66% better (5 to 3) instead of 100% (4 to 2), since humans almost never use conscripts as a fighting force unlike AI.

I have not thought about horsemen, but:
* AI prefer swordsmen to horsemen anyway
* we can increase retreat chances

PS. I do realize it is a bold change that can be considered "outside of the scope of AU", but have not "we" mangled armies?

----------------
About weak militaristic trait:
I think we need to do something to militaristic trait anyway, since we mangled armies, and armies were the main advantage of being militaristic.

Some thing to consider:
* Colliseum could be 120 shields and made militaridtic (other stats like AU mod), so militaristic civs will have them for 60 shield. Roman republic with colliseums everywher
* same as above, but +1 to upkeep and add flag reduce war weariness.
* Police Stations: add flag militaristic, and reduce upkeep to 1 gpt.

[alexman]

Originally posted by pvzh
And more importantly vets only 66% better (5 to 3) instead of 100% (4 to 2)

Those figures are misleading, because they don't translate to odds of victory.

Consider a human veteran Swordsman attacking an AI regular Spearman.

Right now the 4HP Swordsman versus the 3HP Spearman has a 78.7% chance of victory. With your proposed change (5HP Sword vs 4HP Spear), it has a 78.8% chance of victory, which slightly helps the human.

Some thing to consider:
* Colliseum could be 120 shields and made militaridtic (other stats like AU mod), so militaristic civs will have them for 60 shield. Roman republic with colliseums everywher
* same as above, but +1 to upkeep and add flag reduce war weariness.
* Police Stations: add flag militaristic, and reduce upkeep to 1 gpt.

These are good ideas. Maybe we can consider them in another thread, since we have weakened the trait by weakening Armies in the AU mod.

[nbarclay]

The reason why we considered such drastic changes to armies acceptable is that AI use of armies had equally drastic bugs in it. AIs don't use armies at all under the stock rules because they don't build armies with leaders (and remember that the Military Academy requires a victorious army under the stock rules). Further, even with the Military Academy providing free armies periodically (and not requiring a victorious army before it can be built), AI armies can end up sitting in a city with just one unit in them. And even when AIs use armies in spite of those obstacles, the bug in how AIs use armies tends to result in sub-optimal unit mixes such as one infantry and two cavalry (thus not getting either the firepower advantages of an all-infantry army or the movement advantages of an all-cavalry army). Addressing such drastic bugs requires a drastic solution.

In contrast, while the AI tendency to build so many regular units certainly reflects sub-optimal gameplay, it is hard to characterize that behavior as constituting a bug. The advantage human players get from a higher proportion of veterans is merely incremental, not a case of humans' having a weapon so powerful that AIs are completely incapable of coping with it. Indeed, on higher difficulty levels, AI advantages in production costs and free unit support go a long way toward offsetting the human advantage of a higher proportion of veterans.

Also note that although our change to armies was drastic, it affected only a peripheral element of the game. In contrast, adding a hit point across the board would affect every unit in the game, and would make the feel of every single combat different in the AU Mod from what the same combat would be like under the stock rules. Indeed, for players who switch back and forth, AU Mod regulars would look like veterans and AU Mod veterans would look like elites. The potential for confusion inherent in such a situation is more than just minor.

[pvzh]

Consider a human veteran Swordsman attacking an AI regular Spearman.

Alexman, your numbers are wrong you forgot to fortify that regular spearman

Ok. More seriously. I ran numbers through Civ III Combat calculator to get victory odds (see tables below)

1. Archer Rush. Archer rash becomes harder, because archers indeed get reduction in odds by ~ 4 - 5% assuming grassland town (see table 1) and fortified regular spear, and for town on the hill things become nasty ~ 6 - 7.5% (see table 2). But, guess who is doing archer rushes? Not the AI

2. Swordsman. Human swordsmen are either vets or elite and AI defends mostly with regular spears (fortified on grassland: see table 1; or fortified on hill: see table 2). For grassland adopting new system will lead to reduction of swordsman odds by ~ 2 - 3% (grassland) and by ~ 3.5 - 4% (hill).

3. Horseman. Human horsemen are vet and elites as well, and AI does not use horseman much unless it has no iron, but has horses.
Horses suffer considerably in the odds of winning (as bad as ~7%: see tables 1 and 2), but almost all is recovered in increased retreats (not great, but AI does not use horsemen much)

4. AI counter attacks on advancing human swordsman. As slowmover advances in the enemy territory it can be easily counter-attacked by AI archers, and surprise (!), adopting new system will slightly improve odds of AI archers (same will apply to counter-attacking horsemen).

5. Conscripts. The main point of this new system was to improve odds of conscripts, because AI drafts units relentlessly when city is threatened albeight conscripts have low chance to do anything. However, under new system human chances to defeat conscript will drop by ~ 5 - 10% (see table 4).

So, Human tactics suffer more than AI ones

Edit: to Nathan

AU Mod regulars would look like veterans and AU Mod veterans would look like elites. The potential for confusion inherent in such a situation is more than just minor.

Well, indead it might be a problem, but would it be more problem than Light and Heavy Cavalry confusion?

  
  



NB:
number in brackets is % retreated for fast unit

Table 1. Spearman is fortified
and on grassland


  

    

      

      
      
Swordsman
vs
Spearman
      
      
Archer
vs
Spearman
      
      
Horseman 
vs
Spearman
      
    
    

      
Stock
(2/3/4/5)
      
      
New
(3/4/5/6)
      
      
Stock
(2/3/4/5)
      
      
New
(3/4/5/6)
      
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
    
    

      
Vet
vs Reg
(attacker win%)
      
      
70.4%
      
      
69.3%
      
      
50.85%
      
      
46.6%
      
      
48.9%
(25%)
      
      
45.6%
(28.5%)
      
    
    

      
Elite
vs Reg
(attacker win%)
      
      
81.4%
      
      
79.5%
      
      
63.4%
      
      
58.1%
      
      
61.9%
(19.4%)
      
      
57.3%
(24.1%)
      
    
  


Table
2. Spearman is fortified and on hill

  

    

      

      
      
Swordsman
vs
Spearman
      
      
Archer
vs
Spearman
      
      
Horseman 
vs
Spearman
      
    
    

      
Stock
(2/3/4/5)
      
      
New
(3/4/5/6)
      
      
Stock
(2/3/4/5)
      
      
New
(3/4/5/6)
      
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
    
    

      
Vet
vs Reg
(attacker win%)
      
      
58.2%
      
      
54.9%
      
      
38.1%
      
      
32.3%
      
      
36.6%
(31.3%)
      
      
31.7%
(35.9%)
      
    
    

      
Elite
vs Reg
(attacker win%)
      
      
70.6%
      
      
66.5%
      
      
49.9%
      
      
42.5%
      
      
48.6%
(26.6%)
      
      
42%
(33%) 
    
  




NB: advancing
units in the enemy territory is not fortified

Table 3. Regular Archer on
counter-attack vs Swordsman

  

    

      
 
      
Swordsman
(grassland)
      
      
Swordsman
(hill)
      
    
    

      

      
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
    
    

      
Reg vs Vet
(attacker win%)
      
30%
      
      
31.3%
      
      
17.9%
      
      
17.4%
      
    
    

      
Reg vs
Elite
(attacker win%)
      
18.9%
      
      
20.9%
      
      
9.6%
      
      
9.9%
      
    
  


Table 4. Conscript fortified in
city on grassland

  

    

      

      
      
H.Cavalry
vs
Rifleman
      
      
Tank
vs
Infantry
      
    
    

      

      
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
      
Stock
(2/3/4/5)
      
New
(3/4/5/6)
    
    

      
Vet vs
Cons
(attacker win%)
      
55%
(18.9%)
      
      
45.7%
(29.5%)
      
      
74.1%
(10.6%)
      
      
69.4%
(16.3%)
      
    
    

      
Elite vs
Cons
(attacker win%)
      
66.5%
(14%)
      
      
56.35%
(24.9%)
      
      
83.9%
(6.5%)
      
      
79.1%
(11.6%)
      
    
  

[alexman]

What about veteran AI longbowmen versus human Riflemen/Infantry, and other such situations where the AI uses obsolete units, and relies on numbers to win battles?

Anyway, even forgetting about such cases, I think this change would require too much rebalancing to be part of a conservative mod like this one.

[punkbass2000]

"Well, indead it might be a problem, but would it be more problem than Light and Heavy Cavalry confusion?"

Yes, without doubt. What's confusing about L&HC? You're fundamentally changing basics of the game with increased HP. Doubling HP would sound far less impactful and confusing to me, and yet still not really be for this mod, IMO.

[nbarclay]

Originally posted by pvzh

Well, indead it might be a problem, but would it be more problem than Light and Heavy Cavalry confusion?

In my view, yes. The issue of light and heavy cavalry offers a certain amount of potential for confusion during a particular portion of the game, but it has no impact at all before cavalry become available and very little impact after they become obsolete. In contrast, a change in the number of hit points for every unit would affect the entire game from start to finish.

Or looking at it another way, what we've done with cavalry is a carefully tailored change aimed at addressing a specific problem in one speciffic part of the game. In contrast, adding a hit point to all units would shift the balance a bit in combat throughout the game, from start to finish.

[Dry]

If I understand well the problem - conscript-ed units, not barbs or units from huts - it concerns only riflemen, Infantry and MechInf. (the conscriptable units after Nationalism).
What about giving 1 extra HP to those units only.

I think this would also more difficult for the human when using artillery.