SiegeAI,CAI,BAI observations. More observations in post 30 that have been updated.

[Dunadd]

Good post Huberto - i tried to rep you for it but it won't let me (gives a "you must spread some rep around first" message)

[Totalheadache]

@aenima i get where you and Setekh are coming from of course...but no big deal surely?

most of us realise that the patch wasnt specifically addressing sieges...

[AnthoniusII]

Lets put things in a line...
NO one said that patch 13 was expected to fix AI !!!!
Read the OP carefully..
What i say is that I play a campaign in patch 13 (there for latest patch) to see what is new and the same time observe the AI while playing!!!!
When someone posts the number of a patch often means that he/she plays the lastest version of the game.
That helps others-including CA developers- to categorise feedback observations and see what is needed to be fixed!
In nowhere in the OP i say "CA you promised to fix AI in patch 13...liers" or something like that!
In the OP you can find links from older observations inorder to help us all compare things! Why you focus on the patch's number?
What we describe here in not the number of a patch but how the game "behaves"...
If you think that I am too offensive or just biassed over CA and R2TW game please ignore this thread.
As posted hundreds of times in TWC , we do not critise just for critisism...We are trying to focus and reveal issues that need to be fixed inorder ALL OF US will enjoy a game.

[DavidtheDuke]

Lets put things in a line...
NO one said that patch 13 was expected to fix AI !!!!
Read the OP carefully..
What i say is that I play a campaign in patch 13 (there for latest patch) to see what is new and the same time observe the AI while playing!!!!
When someone posts the number of a patch often means that he/she plays the lastest version of the game.
That helps others-including CA developers- to categorise feedback observations and see what is needed to be fixed!
In nowhere in the OP i say "CA you promised to fix AI in patch 13...liers" or something like that!
In the OP you can find links from older observations inorder to help us all compare things! Why you focus on the patch's number?
What we describe here in not the number of a patch but how the game "behaves"...
If you think that I am too offensive or just biassed over CA and R2TW game please ignore this thread.
As posted hundreds of times in TWC , we do not critise just for critisism...We are trying to focus and reveal issues that need to be fixed inorder ALL OF US will enjoy a game.

Agreed. At this point I'm only interested in CA's Warhammer game, and their progress on this game may indicate how far they will get with their Warhammer game. If their Warhammer game comes out and it is horrible too, I may just lose all interest in CA.

If everything *but* the siege AI in the Warhammer game is great, I might buy it at 50% off or something with all the DLCs.

[PeonKing]

I'll admit, I haven't read through the whole thread, but I've actually experienced a pretty huge improvement on siege AI. Basically this is how it went down, I had just taken Athens as the Odrysian kingdom with a single stack that's half thracian warriors half thracian peltasts. (Well, 9 warriors and 1 general, 10 peltasts.) I got attacked by a two Spartan armies, one full stack one that had about 15-17 units, as well as two Athenian ships reinforcing. The full stack was the main siege army, which came with 4 ladders. They manned the ladders, but they actually continued to try to re-man the ladders after I led a sortie to kill their ladder troops. They actually succeeded in landing a set of troops on my walls, took a tower, and ran for a capture point. This was accomplished because the other stack of 15-17 troops was burning the gates on the other side of my city and I had to divert a lot of my peltasts to whittle down that army before breaching the gates. All in all not a terrible strategy by the AI.

There was still some drawbacks. First, the AI which you used to be able to trick to abandon their ladders by attacking them with melee troops, now was only mindlessly pushing the ladders, not trying to defend themselves from my attacking Thracian warriors even though I only led 3 sets of units out to attack 4 groups of ladder troops that weren't far apart from eachother. Secondly, they didn't attempt to engage my sortie troops with the bulk of their army, despite these troops not even being Cavalry and I led them far from my gates, they could've easily routed those troops as they had 2 units of Companion Cavalry mercenaries. Additionally, they used a pretty crap strategy to try to burn my gates down, since if they had thrown more troops at the gates before I got my Peltasts in place to try to repel them, they probably would've breached the city earlier and been able to overwhelm my troops.

All in all, I see some significant improvement in siege AI strategy, but it needs risk management improvement.

[DavidtheDuke]

I'll admit, I haven't read through the whole thread, but I've actually experienced a pretty huge improvement on siege AI. Basically this is how it went down, I had just taken Athens as the Odrysian kingdom with a single stack that's half thracian warriors half thracian peltasts. (Well, 9 warriors and 1 general, 10 peltasts.) I got attacked by a two Spartan armies, one full stack one that had about 15-17 units, as well as two Athenian ships reinforcing. The full stack was the main siege army, which came with 4 ladders. They manned the ladders, but they actually continued to try to re-man the ladders after I led a sortie to kill their ladder troops. They actually succeeded in landing a set of troops on my walls, took a tower, and ran for a capture point. This was accomplished because the other stack of 15-17 troops was burning the gates on the other side of my city and I had to divert a lot of my peltasts to whittle down that army before breaching the gates. All in all not a terrible strategy by the AI.

There was still some drawbacks. First, the AI which you used to be able to trick to abandon their ladders by attacking them with melee troops, now was only mindlessly pushing the ladders, not trying to defend themselves from my attacking Thracian warriors even though I only led 3 sets of units out to attack 4 groups of ladder troops that weren't far apart from eachother. Secondly, they didn't attempt to engage my sortie troops with the bulk of their army, despite these troops not even being Cavalry and I led them far from my gates, they could've easily routed those troops as they had 2 units of Companion Cavalry mercenaries. Additionally, they used a pretty crap strategy to try to burn my gates down, since if they had thrown more troops at the gates before I got my Peltasts in place to try to repel them, they probably would've breached the city earlier and been able to overwhelm my troops.

All in all, I see some significant improvement in siege AI strategy, but it needs risk management improvement.

That's great. I hope to see more stories like yours, though almost everyone is experiencing not-that right now, unfortunately. I hope they fix it before the Warhammer TW (if it is TW) comes out.

[riskymonk]

I believe sega is starting to see CA as a liability as reputation plummets and profits do not meet expectations. The contract may end spelling the end of CA. Most people blame sega however CA have their own development and marketing team so it's not all Segas fault.

[AnthoniusII]

IF CA will drop any more attempt for fixing this game to a descent level will risk to lose atleast 25% of its customers for its next product!
That 25% is the TW hardcore players that buy almost all TW games. 25 may look small as a number but it is a huge one in real numbers of customers and there for money!!!!
I hope that CA wont try this way for this game!

[PeonKing]

IF CA will drop any more attempt for fixing this game to a descent level will risk to lose atleast 25% of its customers for its next product!
That 25% is the TW hardcore players that buy almost all TW games. 25 may look small as a number but it is a huge one in real numbers of customers and there for money!!!!
I hope that CA wont try this way for this game!

Where did you get your stats from?

Furthermore, I disagree, I am happy they have made progress, as shown in my previous post. The game is significantly more difficult now than it was a few patches ago, I'm glad they are talking time to fix up their product.
I hope they continue to release patches, maybe some with new battle maps like they did with S2.

[jtfjtfjtf]

I'd love if some infantry could scale walls. But the siege and small town AI is still poor. In small towns it still just sends everyone to one opening instead of exploiting multiple openings.

[Sidious]

Can't they make Seige AI scripted in stages?

do a quick check if siege equipment exists if so then burning gates=null

stage 1, move ladders to walls,
stage 2 move battering ram/use siege equipment
if x amount of men die before walls or gate taken then rout,
if not push threw.
then do another quick check if siege equipment exists if so then burning gates=null if not either rout or attempt to burn gate.
Rinse and repeat,

[jtfjtfjtf]

Can't they make Seige AI scripted in stages?

do a quick check if siege equipment exists if so then burning gates=null

stage 1, move ladders to walls,
stage 2 move battering ram/use siege equipment
if x amount of men die before walls or gate taken then rout,
if not push threw.
then do another quick check if siege equipment exists if so then burning gates=null if not either rout or attempt to burn gate.
Rinse and repeat,

I feel that the AI should take its time building up before sieging. Usually they instantly assault the city so they have no battering ram to begin with, just the 4 ladders.

[14182]

I feel that the AI should take its time building up before sieging. Usually they instantly assault the city so they have no battering ram to begin with, just the 4 ladders.

The attacking AI does waits and builds extra siege equipments when certain conditions are met, such as not significantly outnumber, no other field army from the defending faction can reach the region in x turns.

But it's a paradox since the CAI is quiet passive and will only attack a player's settlement with a great number advantage.

[chris10]

Besides the fact that many people dont come here anymore on a regular base (Iam one of them), cause of the constant trench warfare with certain individuals who downplay or outright try to talk away major issues with the game, its frustrating to see that after 13 patches a key feature of ancient warfare is almost as broken and utterly annoying as it was on release but there was always manpower available to produce a busload of (cashmaking)dlc.

Its honestly pathetic and those of you guys who still think that CA will really fix the sieges after they havent been fixed until now are clutching at straws...its about time you accept reality...it wont get fixed cause it cant get fixed as the current AI design and its code would not support or is incompatible with the neccessary changes.

Knowing basic AI programming I say that if sieges could have been fixed they would have been fixed a while ago and CA is pussyfooting on that issue cause it knows the sieges can not be repaired in the current code design without seriously breaking something else so unless they are not secretly working on a new AI to patch it in secretly on a later stage (seriously, what are the chances for that ?)...well...this is what you get.



The last one turns off the light ,pls

[AnthoniusII]

CAI,BAI,SAI are that bad that i just finished my Rome campaign in HARD dificulty..and I feel that the game-as is now- does not worth any more of my time.
I hope that impresion will be changed in a possible future fix....All that remains for me is the existance of those wonderfull horses...
Strange...TW games always lacked realistic horses but offered an average game chalenge...Now R2 reversed that...Wonderfull horses but no game chalenge...
I hope...still hope!

[Turbo]

I wonder if CA is going to use the Paradox model in disguising a patch as an expansion. It doesn't look like CA is going to fix the AI with patches at this point. I suspect that there is too much wrong under the hood for a normal patch to fix. Perhaps they see a financial incentive to make some more money and finally fix Rome 2. The only alternative I see is a complete abandonment of Rome 2 once they have exhausted the DLC.

[Yerevan]

Players defending R2 through only rethorical questions is something I see more and more around here. To the point I start to wonder if they ever take some time to play this game they enjoy so much.

And if they do like this game, why is it so intolerable to them that other players come here with constructive critics.

Anyway, about the AI concentrating its armies around its capital its something I saw a lot last time I tried to play R2. There's some logic in it : not in the purpose of defending its territory, obviously, but to make sure it will always fight with a numeric advantage.

[Junaidi83 de Bodemloze]

guys relax the rome 2 will up and fit for play maybe after this x mass, provided that CA will work for another new expansion, as long the dlc come up the future patch will come,
most of total war game always stable after 1 year release (stable = more better then day 1 release )

Hahahah if Darth still active modding, i already know what will follow (Do you see what i told you CA .....bla...bla..bla) Well at least he got the point. Just look around many mod from previous total war still up and running for you war monger fan base.

[ABH2]

In such cases AI army tried to find high ground to defend better. In cases that the number of units -despite the diference in quality- AI simply waits the player to have the total control of the battlefield. I am playing in Hard and Very Hard dificulty.

This has, from my experience, always been the case unless the player has siege weapons in their army. When they defend, they sit in place until reinforcements arrive. So, I don't see anything different here in patch 13.

In few cases AI starts ONLY with one ladder -instead of 4- espesialy when the settelent that defends is level 2!

I haven't had time to play a lot with the new update, but the only instance in the past in which the AI had one latter from my experience was when it was a rebel army. That was the case in previous patches.

AI still uses torches only to destroy gates. Ussualy attacks another gate again with suicide attempts!

Well, maybe it's just me, but I consider the AI attacking two gates in itself progress. Is it a great leap forward? No, but it's better than cramming every single unit into the same gate.

[AnthoniusII]

Some updates:
This time exept reinforcements we will see C(ampaign)AI and few observations on B(attle)AI as well.
I have good news and bad news as always.
The good news are that durring a multi armies battle AI found a "narrow" path behind my lines and sent its cavalry in my rear.
I must admit that i did not expect to see that in patch 13 but i did and it was a pleasant surprice!
Bravo in this point CA devs.
Now I will show you some campaign pictures and few battle ones to point issues about the game that ergently need fixing.
Lets start with siege AI behaviur.


I show this picture to help you realise the chalenge of my battle that i will describe:
Lets's start.
Civil War Event and Senate Loyalists apear in a huge urban center (this time Athens while in a previus campaign of mine was Massalia).
6 of my legions were close enough to seal the area and provoke battles in chock points that despite my defeats did not allow Rebels to replenish fast their forces.
Their fleets left Athens (this is crusial) and I found a small oportunity window to re capture Athens driving Rebels to hanger.
In the battles i had in those armies re-inforcement were just like i described in my OP post. Two armies in each time start the battle and new single units enter the battlefield when a unit stop to exist!
Now this is most interesting.
The 6 fleets of Rebel Romans were intact. They sent two armies (with less that half of their stregnth comparing to the strength they had when they spwaned (see pic above).
I have two legions with huge casualties in every unit of theirs after the bloody recapture of Athens. One as guarison and one as reinforcements. Athens has actually no city guards because is still one tern after the city recapture and those units have max 10 warriors!
The two legions attack from the same direction , something i was expecting....
Three full stack fleets are attacking from the sea.
I was expecting attack from all long line wall sides of the city but....
Two ships syrpiced me pleasantly when they entered directly in to Athens port!!!! That made me to expect a real chalenge that would made a defeat enjoyble!
Instead all the rest of the ships disembarged their troops in ONE shore ONLY!


Even in this case what you can see in this picture besides the impresive visual result of all these ships is that:
1: Despite the fact that i sent almost all my units to defend that gate infront of that invasion force my units were small in number (see my cavalry units).
AI had a number superiority 30/1 when 3 units of those burned that gate!
And when that gate was destroyed AI desided to lead all that force to the gate the two legions tryed to destroy (nottice in the minimap where are the siege engines).
In their journey to that destination they lost 1/3 of their forces by the wall towers.
In the minimap you can see that AI did not attack in the city's right gate (as we see it in minimap) at all!
Despite the fact that AI had more than 70 units in total chosed to attack in one side ONLY even when it HAD a second gate destroyed as well!
Remember that i play in HARD difficulty.
CA must forcus on players and AI reinforcements options plus to "teach" AI to cover more areas of assault like in RTW1.
AI siege forces must be able to wait atleast 1 turn without atrition penalty inorder to deploy more siege machines. Even if AI will continue use torches then allow it to use Gallery as protection of those units.

Now some CAI issues:
Few may remember that i posted in TWC and as well in CA official forum about the AI senceless naval journeys in patch 4 or 5.
CA did fix that issue and CAI is quite better than that level.
Few issues remain though to be fixed.


Is well known that Sparta in most cases (and Athens in less ones) are quite OP thanks to CA's choice to give those factions clasic Hoplites and make them to much stronger that other hellenistic factions infantry.
But this is not the case of this picture. Right or wrong what we can see in this picture is a list of facts:
Player's faction (Rome) has reached to a point.
In the east there nottice that a faction (in this case Sparta) has expanded from Great Germania to Sarmatia and entire eastern europe.
So far so good! But...Nottice that in easter mediteranean sea there is another huge power (in this case Ptolemies) and here were bizzare things start.
Ptolemies are CLIENT STATE of Sparta and became client state when THEY were ALREADY a huge power! Strange indeed that CA must look at it asap!
Now please see the following picture.


You can see players offesnive against the east super power!
The small minor factions that shown in the map are results of invasion of the player and then leaving settlements to rebel.
Here start the new bizzare issues.
1: AI instead of maintain forces in several regions (atleast capitals of provinces) tent to gather atleast half of its forces arround to its capital leaving the rest of the state totaly ungaurded.
Exibit 1:


When war between Rome (player) and Sparta (AI) started , both had 10 armies each! Please count how many armies are arround Sparta and compare this pic with the extend the Spartan state had at that momment!
More bizzare issues! Sparta kept intact only half of those armies and left the other half with half of their stregth (i never understood why). I was prepaired to deal with those 6 armies and what i found was Spartan generals with 5-6 units each trying to reach spartan cities.
Bizzare issues continue.
In my offensive i found NO OTHER spartan armies expept those ones!!!!!!!!! Where were the rest????????

To prove the fact that all AI factions protect ONLY their capital:


So far CAI has been improved by CA comparing to patches like no 4/5.
What CA must continue to improve are the following:
Ballanced AI building constructions to allow AI :
To have and maintain the maximum armies an fleets available to its size!
To allow armies survive of hanger if a small number of provinces will lost.
To achieve that regions must have a ballnced food/religion/millitary etc constructions.
So far i saw AI armies with several type of soldiers and that is very possitive!
AI must be able to chose strategic points of defence and rulling is own states!
Ofcourse capital has more value but AI must:
Have armies in the 4 sides of horizon to defend its borders.
Must be able to keep a strategic reserve (1/4 or 1/6 of total armies and fleets) to commint invasions in spesific areas.
For example lets supose that AI Carthage must invade Gual.
Carthage must keep armies in the eastern borders (africa and sicely) to defend.
Cartahge must keep armies to defend Carthage its self and keep forces to maintain peace in spain. In that case two full armies and a fleet must be the invasion forces.
Unless an AI faction has no enemies in one side of the map (se sparta in our example) then AI must enforce the other borders or add more invasion forces!
Finally what i naver found in my campaigns is the betrayal by AI "allies" even when vast areas of mine have no armies to defend them!
See my pictures above:
Carthage and its client states are allies of mine.
But before make them allies by force i took several regions of theirs in multiple wars!
In RTW but strangely in M1TW when i moved my legions to the spartan front Carthage would betray me!
Where is this feature anyway???
EDIT: update...
Continuing about CAI ...
It seams that AI factions can not hundle properly the building constructions.
CA made some improvements but still....
Continuing my Roman campaign i notticed that Rhodes attacked Ptolemies in Asia Minor. Knowing about the "habbit" of AI to gather half of its armies around its capital at first did not look bizzare!
But i notticed the fast advance of Rhodes in Asia minor in only 3 turns!


So i desided to invastigate the matter.
I attacked in Ammaseia that had no guarisson atall because of hanger!
I moved my spy and see what it found!


Food is a good as factor is a good addition in game i must admit. But is a knife with two edges!!! Its not chalenging in hard difficulty to fight this kind of armies isn't it????
In my humble oppinion the solutions are simple to be in the next patch.
1: Increase by 100% or more the available slots of buildings in each settlement (12 in province capitals and 8 in each provincial city).
That way AI will have more options to have both survival and millitary focus on its building choices.
2: Addition of province infrastructure building slots like ETW.
While AI or player will have an entire province under his rule 3 slots of available buildings could be available like Roads, borders forts or anything will come to mind in CA developers.
3: Reduse the cost in food by 50% in spesific buildings or remove that cost entirely for some of them.
That way AI will spend also less FOOD to its attempt to develop.
In any case AI must be able:
To survive no matter how many and which provinces will have in its disposal.
To develop its millitary might and create "realistic" armies and fleets.
To be able to keep on track techonogical advances inorder to field up to date armies and also be able to keep its state inorder.