AI of defending army is quick to leave their hexes where they have defense buffs and walls to attack outside of the walls

Now this can be intentional if:

1. The defenders have no chance of winning, no reinforcements coming and enemy has range units and/or siege equipment: sortying out can provide max damage to the attacked in that case

2. The attacker has exposed themselves in a very vulnerable postion: low elevation, river crossing and thus the defender has a chance to inflict higher damage then waiting behind the wall

The above are valid cases but I had the sense that the defending AI is always leaving their position if they can

Sometimes they cant as my troops are waiting just outside the wall but then they attack those units: wouldnt they inflict more damage if they wait for me to attack?

That said I always have like 50% range units so the AI might rightfully evaluate that waiting behind the wall is a sure way to losse health without inflicting damage itself which brings us to:

TIME: the most important asset of city defenders in any era and with the warscore system it is more relevant then in other 4x games

It needs to be evaluated that fortifications were there to win time not to win battles. 

Thus attacking from a defended position would only make sense in the above 2 points. In other cases the defenders are playing on gaining time for reinforcements to arrive or to simply tie down as much of the enemy force as long as possible

In that vein the defending AI might need to hold the best defendable positions until its health runs out or an opportunity presents itself to inflict max damage at an exposed attacker unit.

This is especially true since in this game the warscore ticks away by loosing battles and loosing citites: so holding on as long as possible provides valuable time to turn a war

On the flip side: If the AI is dragging out the siege, some players might be annoyed at how passive it is

But I think its reasonable to assume that the players will understand the desire of the defender to not attack from fortified position and that we need to crack these nuts ourselves to get quick results

Could militia be given an ability that they can a ranged 2 attack from a garrison? Probably not. Unit types.

Daarkarr0w wrote:

Hi @Zolobolo,

Do you have any save for the match? That would help the team to analyse the scenario and replicate it.

Please find a save below:

Siege AI.ctr

This save is suitable to test a number of scenarios described below for the following reason:

1. It is taken directly after I ahve initiated a siege on an enemy city

2. No reinforcements available on either side

3. No range units available on either side (made sure to attack wth swords and spears to get clean results)

4. Besieger is in a clear advantage (it would be prudent by the defenders to attack if they had the advantage in strenght but this is not the case here)

5. City is located in an optimal location to hold up for a long time as there is only a single-hex route leading into the city and the city walls can only be attacked from a single hex. On top of that the city also has elevation within it enabling the defenders to hold out even if the wall is breached

Case #1:

The defenders decide to attack the besieging force.

This is a clear-cut wrong decision to take as they are:

1. At a disadvantage in overall strenght

2. The landscape puts them at a disadvantage when attacking (as they would need to both leave their walls and attack uphill)

3. There is likely a reinforcement not far away and/or one could be generated fairly easily to relieve the siege

Conclusion: Should not attack but wait at least till a siege engien has been finished but likely even beyond that

Case #2:

If attacker leaves the hex in front of the wall empty, a single defender unit moves in there in Turn 1

The unit does not attack afterwrards which is good but is in a location where it will be wiped with a single charge and causing minimal to no losses to the attacker

There seems to be a function which seems to deintify this situation as no further units are moved into its location one the unit is wiped in front of the walls

There is a chance that the intial unit is only moved in that exposed location outside the walls as there are 4 defending units but only 3 city hexes to place them on. Thus the AI script might have decided to move the unit there jsut to be able ot spawn the 4th defender

Conclusion: moving into the exposed hex in front of the walls is a clearly wrong decision as it neither offers an oportunity to attack but is sure to end in a unit wipe with no damage caused to the attackers. If this calculation is overwritten by the script just so that an additional unit can be spawned, that shoul be fixed as it is better to have a unit in reserve then sacraficing another unit to get it into map as the overal lsum of the decision is one unit less

Case #3:

The other defending units are holding their position when exposed unit is lost or enemy is in front of the walls

This is the corect behavior

Conclusion: there is obviously a correct potential damage calculation made efore unit positioning but it seems to be overritten in some cases

Theory:

A lot of the attacks fro mmy past sieges might have had something to do with my range units being there: maybe the defenders deem the damage they cna caue to archers more valuable then the damage reduction they recieve from the foritifcation and maybe tehy even ignore that handgun units do not get a debuff when attacked - This theory needs another save game scenario to test out

AI Recommendations:

1. When besieged, the AI should not counter attack unless relative strength is at minimum 2:1 in its favor

Rationale:

- Even if the ration is 1:1 attacker is always at a disadvatange (just like in real life) - thus the realtive strenght needs to be clearly better to pull off an offense successfully

- Primary goal of the defender is to gain as much time as possible for reinforcements to arrive and attack the besiegers in the back. Once surrounded the besiegers are in a clear tactical disadvantage from the getgo and levies are valuable in that they represent pops that are lost wth each draftee

- Secondary goal is to inflict as much damage as possible to the besiegers. This can only be done when defending as the fortification bonus is only applied then

- After the AI decides to sally out and counter attack the besiegers, it correctly calculates that it should not actually attack and this is idle: this is annoying for players who should be defending in this scenario and will likely simply skip each turn but its also counterproductive to the AI as it will loose the battle without inflicting any damage once the max turn counter runs out

2. When combat begins, the defenders should not leave their fortified hexes to less secure positions even if thee are still units in reserve that cannot enter the map due to this. They can leave their fortified position if there is a hex that provides even greater avatage due to being higher elevation or enemy units crossing rivers around it

Rationale:

- Levies are valuable and must not be thrown away easily

- Defenders excercise maximum damage_given/damag_taken ratio when defending in fortified hex (except when attacked by archers or siege equipment but well get to that in antoehr test)

Balancing Recommendations:

1. Levies should always be ranged units so that they can always dish out damage to attackers even if they cannot leave their walls without loosing the advantage of being on defense in a fortified position. If this is not possible, then give town watch and following bonus effect to camps family infra (such as Watchtover): Turns +1 levy unit into a Levy archer during siege. The workaround would be fine for the first 3 eras but is a bit more complex when we get to gunpowder units as the benefit of a ranged levy would be gone and another benefit would be needed to repalce it

2. Walls (not fortification hexes in general) should provide protection from close in combat to ranged units (and thus levies as per above point). This is to emulate that the archers are located on top of walls and cannot be easily rolled over via shock troops

Ningauble wrote:

I kind of appreciate that you should plan ahead but that's not a bad point. Archery or a culture's equivalent wasn't uncommon. It did require a bit of specialized equipment though maybe one archer per 2 melee militia? Not half and half.

Yes a 50-50% split was my initial thought as well

Only downside is that its difficult to clearly communicate this rule to the player as its not that intuitive to do

So what might be better is to provide the long range bonus to levies via infra: each infra in the family of camp could increase the number of levies upgraded to archer levies and this fact and the exact amount clearly communicated within the desription of the infrastructure improvement itself

Course there would till be a minor issue: how to clearly communicate the amount of levies and levy archers to the player before they besiege a city

HumanKid wrote:

i feel that the lack of ranged defenders has made melee levies and such become over-buffed... i don't think it is "fair" that urban recruits are made to be as strong as trained forces

if a faction needs a strong defense, then a proper army should be raised

Levies are strong but they kinda need to be as:

1. Players would auto-resolve and instant-capture cities

2. Due to auto reinforcemment in captured territory, the above army wouldd auto fill up soon further increasing the chance for the player to not actually play sieges

3. Due to the high upkeep cost not all citites can be defended

4. Its rather boring for a player to organise tatic army defenses for cities

My argument is actually going to other direction: increasig the power level of levy defenses by making them ranged and altering city wall effect to negate the hand to hand combat penalty to these archer levies

The goal is to have AI cities better defend against both other AIs but particularly against the player so that the foritfications can play the role they ment to play: as a road block

It should take time to build up siege equipment and aso have some levies starve before a fortification can be captured (even without army to defend it)

Yes the AI shouldd train armies but should use them to relieve its besieged fortifications - its iportant for this mechanic though that the fortifications cannot be captured right off the bat 

Currently we can pretty safely roll over a city with roughly equial force protecting it as the levies (though strong) cannot make use of their true strength: the fortification when attacking

HumanKid wrote:

you brought autoresolve up and all it was said is that it should be worked on its own. i won't discuss it any further

Yes I did, this is the argument thus far:

1. Problem: AI controlled cities are frequently rolled over in a singleturn without having to besiege the city

2. Reason: AI sortie out even though they only have levies and then often take uboptimal position

3. Proposed solution: Prevent AI to sortie out unless in clear strenght advantage and adjsut tactical AI to not leave the fortification unless clearly beneficial opportunity is presented

4. Proposal (from you): Reduce strenght of Levies as they are to strong compared to regular troops

5. Counter to #5: This change for #4 would compound the issue mentioned in #1: If levies are weaker they will be even less able to hold off an attack evne from fortified position

6. Additional argument presented to #5: If overall defense strenght of cities is reduced by #4, then we could autoresolve the battles and not even have to command them (besides not having to siege) so a new problem is generated

7. Counter to #6: If autoresolve is deemed problematic, then it should be addressed on its own

8. Counter to #7: Autoresolve is not the issue (it should yield the same result as both sides played by the AI - but cant say as I never use the function). The point made in #6 was that reducing the overall strengh level of levies would make them so weak that we wouldnt even need to play the battels to roll them over at turn 0 of a siege which would lead to autoresolve sieges as a foregone conclusion

Summary: Auto-resolve was brought in as an argument agaisnt a change proposed. It is not an issue right now (to that I would know if anyhow) but it would become an issue if change implemented. An issue that could not be resovled by tweaking the resolve formula (as it should not be tweaked it should jsut be a sped up AIvsAI simulation)

You have also made this argument:

"i feel that the lack of ranged defenders has made melee levies and such become over-buffed... i don't think it is "fair" that urban recruits are made to be as strong as trained forces "

I havent played a state of the game where levies were ranged so cant say if that is true

I agree that levies are strong compared to regular troops but I belive they need to be in order to have any sort of chance to hold out for at least one turn and even now they are still too weak due to being mele :)

Would instead turn them into ranged units: less powerfull then professional archers but this time that woudl not be an issue as ranged units can fire from foritifed position (so they dont need to leave their defense) and can pepper the enemy even if not adjacent to them which is super helpful for defendig

I have also argued to ensure that walls negate close range combat penalty of the defenders (emulating that they are high up on secured walls and cannot be easily charged)

My overall consideration is this:

• Foritified positions (cities with walls) are there to be a speed bump to enemy forces no matter the size meaning they need to be able to hold out for at least 1 whole turn agaisnt any enemy force (impossible to defeat at turn 0 of the siege)
• I dont want to bring up real-world examples and considerations as we could do that for any design
• I would argue hovewer that the goal of fortifications should be to act as a force-multiplier: multiplying the force of the (weak) levy defenders to a point where they have to be worn down through siege
• The siege can starve levies out (invest time), build siege equipment (to bring down the wall) and even chop forests to speed up the building of said equipment

HumanKid wrote:
units in every city is overkill. a competent player would place forces only on positions that indicate a possible threat, like a city bordering on an aggressive opponent

please try to argue your points without presenting a single option as the only valid one

Yes you are right: this should not be a considerable micro for a player I take that argument back sicne we tend to have few cities and evne less exposed ones

I didnt want to be disrespectful: I am sorry if I come across as too aggressive

I argue for the points I make and counter to points that challenge them: I could argue also against my own points but dont have that sort of capacity + my posts would be even longer and no-one is reading through these at this point anyhow :)

HumanKid wrote:

we agree that positional combat is the issue. if they are to be just buffed again, that won't solve the underlying problem

i'm also in favor of some form of short ranged attack. i would prefer something other than archers, as they are units to be trained... perhaps citizen defense could trow rocks, boiling oil, whatever... as they actually did IRL

levies do hold on their own on actual gameplay scenarios provided the invading forces aren't overwhelming. that makes sense. they're not armies. just a bunch of peasants with tools for weapons

they could be better at holding position and only do sorties when having vailable reinforcements

Yes we agree that there is a problem

I think that both the AI and the units need to be fixed yes

Fixing the AI is no question - it needs to be done

Rebalancing of the unit is just an additional measure I think is needed as further action is required to prevetn rolling over a city at turn0 of a siege. The goal could be achieved differently of course:

1. Minimum siege time before attack is allowed: would feel too restrictive for players

2. Increase strenght of levies: I agree with you that they should not be made stronger

etc...

Turning them into archers would fit the scenario: they start in limited area (no room for manuever) which is well defended and need to dish out damage without being damaged themselves and even if not next to the attacker

Yes we can call them other then archers but with ranged attack: ideally they woudl need a range of 2 to be able to dela damage even from behind their frontline

Yes they could and should be much better

At least in my own experience playing normal difficulty I can capture a town with 4-5 units which sohuld coutn as a medium sized force

This is partially due to them leaving heir position but mostly due to them havignto attack my mele unit to deal damage (hence my ranged levy suggestion).

All I have to do is breach a walled sector and then place my strongest unit there: once they attack that my unit is fortiifed but they are not and they wear themselves down)

Yes they are just peasant (in this game) but the overall design should dictate what they are how they fight and how powerfull they are

If we agree that they should be able to hold out for at least 1 turn then they need to be much more powerfull in that particular scenario (but not in general when compared to regular troops)

Zolobolo wrote:

Daarkarr0w wrote:

Hi @Zolobolo,

Do you have any save for the match? That would help the team to analyse the scenario and replicate it.

The save might also be good to analyse a fourth case:

Siege AI.ctr

Case #4:

AI does not send reinfocements

The save migth also be suitable for testing AI reinforcement routines as:

1. If AI is not exploited (unit is placed in front of the wall so the AI cannot move in there and loose one unit in a single attack) then the battle lasts for 2 full turns: 2x3 rounds

2. The main city of the besieged faction seems like its 1 turns away with 6 pops thus in theory they should be able to crank out reinforcements (1-2 units) and send them to releave the siege

3. Tested with waiting the full 3 turns as well (max number of turns before I automatically win due to being the defender) and I did see a cavalry unit that seems to have bee ntrying to relieve the siege

The save is not suitable for case #4 if the AI lacks resources to produce (more) reinforcements or there is a valid economical or strategic reason not to do so

e.g.: if they have overal low pop, money and the capital caity is much better defendable

Naturally we also dont want the AI to cheat and get free units nor to bankrupt itself relieving a city it will most ceartanly loose instead of concentrate protection to its better foritifed cities

Daarkarr0w wrote:

Hi @Zolobolo,

Do you have any save for the match? That would help the team to analyse the scenario and replicate it.

Another save for a new scenario I havent see before:

Siege AI 3.ctr

Case #6:

Defending AI decides to coutner attack besieging force evne though their army strenght comes mostly from ships that cannot engage the enemy due to terrain and even if they could most of them cant attack land units

The core issue here is of course the unlucky tactical map that prevetns the denfending ships to engage the enemy at all but there is an AI aspect to this:

1. AI should not rely on strenght that comes from naval ships than cannot engage groudn targets (I have also recommended to allow ALL naval units to engage land units to resolve this exact issue)

2. AI should only count a fraction of its naval strenght when deciding to engage on ground as ships have range limit that can be easily exploited by a ground army

Example: Defending AI has 6 Quadrireme (31 strenght each) and 3 Carracks (41 strenght each)

When calculating its offensive power it should count = 0*31 + 3*41/2

0*31 as Quadrireme is unable to engage land units

3*41/2 as Carracks can engage but are naval units in a land battle and will thus be not to effective

A case could be made that ships with no land attack should also be calculated just with evne less weight e.g.: 6*31/10 as they could engage enemy ships or land units of the enemy if they enbark onto a sea tile