Maddening Army AI (this army was sponsored by RAID)

Users who are viewing this thread

Yertyl

Veteran
I assume we have all seen some idiotic army behavior, but I wanted to share two particularly egregious examples I came across yesterday:
96BD169F2301D0F778C0B9BFD184D511C36F3294


You may notice that a) Our city Car Banseth is under siege b) Panalea's army containing large parts of the troops of my small kingdom is in the middle of enemy territory. What are they doing? Some kind of clever counter-attack to force the army besieging Car Banseth to retreat? Nope, raiding a village. In the middle of enemy territory. Needless to say they all died about 30 seconds later, and Car Banseth was easily taken.
DF91725E1C2DDD68ACB6B74EAACEED8838CD20A2

Here we finally have a realistic chance of taking Dunglanys. So what is my second army doing directly next to me while the enemy is slowly gathering enough forces to crush me? Correct, raiding a village.
I reloaded after this debacle, and the second time, the Battanians went for the Panalea's army instead of me. They easily could have run, but instead kept raiding and all died. Because apparently the only thing sweeter than life itself is raiding villages.
Now, I specifically founded my own kingdom to avoid this AI idiocy. And you theoretically have the option to disband an army, right? But no, for some reason not while they are raiding, because the only duty more holy than your king's word is raiding frickin' villages.

So yeah, currently the complete idiocy of the AI, combined with game rules that artificially prevent you from preserving their lives -- even when they are your direct subjects -- is the only aspect of the game that's frustrating enough for me to consider quitting. The late game experience of watching your allies or subjects take most of the troops of your kingdom to suicide brainlessly or do nothing at all is just painful. And that the game rules actively work against you, like being unable to command troops while they are raiding, or continually eating relationship hits for the raids your brainless compatriots do constantly against your will, or taking relationship hits with every single member of an army for disbanding it when it does something completely idiotic, is the cherry on top.
 
Last edited:
Yes, it's pretty ridiculous the amount babysitting you have to do when being the leader. It's especially frustrating when you have lords obsessively raiding since raiding gives you next to nothing in terms of money anyways.

The ordering of your vassals/clan system cannot come soon enough imo.
 
Ai logic:

Army size < nº of troops needed for siege? Raid.
Army size > nº of troops needed for siege? Siege.

Nothing more and nothing less. It doesn't care about entering enemy territory, it isn't aware of the surroundings until it's too late. It's quite fun when a faction is outnumbered 10/1 and their armies go willy-nilly into enemy territory to get wiped again, and again, and again.
 
Ai logic:

Army size < nº of troops needed for siege? Raid.
Army size > nº of troops needed for siege? Siege.

Nothing more and nothing less. It doesn't care about entering enemy territory, it isn't aware of the surroundings until it's too late. It's quite fun when a faction is outnumbered 10/1 and their armies go willy-nilly into enemy territory to get wiped again, and again, and again.
Well if they changed that to
Army size > nº of troops needed for siege? Siege.
Army size < nº of troops needed for siege? Disband.
It would already be a huge improvement.

I am not sure to which degree it is already supposed to be that way, but armies should be formed
a) to fight enemy armies and
b) to take enemy cities.
and be disbanded when they cannot do either of those things efficiently.
As it is, it seems that AI lords are extremely enthusiastic about creating armies, and then extremely clueless about actually doing anything with them. You do not need an army to raid a single village. It just shoots your own faction in the foot by binding troops that could be used better elsewhere.

The AI should have a clear goal in mind before creating an army, create that army close to that goal, and disband it once there are no more good targets for that army. Instead of faffing about until food and influence run out. Of course, then we might also want to change influence cost to be continual instead of mostly initial, but that would be a huge improvement over constantly having huge armies running around doing nothing but starving themselves out.
 
Last edited:
Yes, it's pretty ridiculous the amount babysitting you have to do when being the leader. It's especially frustrating when you have lords obsessively raiding since raiding gives you next to nothing in terms of money anyways.

The ordering of your vassals/clan system cannot come soon enough imo.
Or when you run around with your companion party and you have given him pretty good troops and he donates some of them to X town/castle garrisson. Or "Hey I just want to check the quest in this villiage... why are you recruiting Aserai peasants when we are Khuzait?"
 
Yea it definitly need some work. The way it is now i try to prevent others in my kingdom from forming armies because they mess up so bad every single time.
 
This is worse than warband. At least its logic made some semblance of reason at times.
Basically it feels like they learned nothing from the frustrating **** that Warband had in it, it's why I can't stand to play a campaign past the first few settlements taken, it's just a meatgrind that YOU personally have to carry out.
 
I assume we have all seen some idiotic army behavior, but I wanted to share two particularly egregious examples I came across yesterday:
96BD169F2301D0F778C0B9BFD184D511C36F3294


You may notice that a) Our city Car Banseth is under siege b) Panalea's army containing large parts of the troops of my small kingdom is in the middle of enemy territory. What are they doing? Some kind of clever counter-attack to force the army besieging Car Banseth to retreat? Nope, raiding a village. In the middle of enemy territory. Needless to say they all died about 30 seconds later, and Car Banseth was easily taken.
DF91725E1C2DDD68ACB6B74EAACEED8838CD20A2

Here we finally have a realistic chance of taking Dunglanys. So what is my second army doing directly next to me while the enemy is slowly gathering enough forces to crush me? Correct, raiding a village.
I reloaded after this debacle, and the second time, the Battanians went for the Panalea's army instead of me. They easily could have run, but instead kept raiding and all died. Because apparently the only thing sweeter than life itself is raiding villages.
Now, I specifically founded my own kingdom to avoid this AI idiocy. And you theoretically have the option to disband an army, right? But no, for some reason not while they are raiding, because the only duty more holy than your king's word is raiding frickin' villages.

So yeah, currently the complete idiocy of the AI, combined with game rules that artificially prevent you from preserving their lives -- even when they are your direct subjects -- is the only aspect of the game that's frustrating enough for me to consider quitting. The late game experience of watching your allies or subjects take most of the troops of your kingdom to suicide brainlessly or do nothing at all is just painful. And that the game rules actively work against you, like being unable to command troops while they are raiding, or continually eating relationship hits for the raids your brainless compatriots do constantly against your will, or taking relationship hits with every single member of an army for disbanding it when it does something completely idiotic, is the cherry on top.

Example 1 : We do not know how many troops is besieging Car Banseth and how many troops Sturgian army raiding enemy village has. Without these datas you cannot say something. Maybe siegers are 3x of army raiding village and they see no chance maybe if they travel there they will just watch siege. They do not know your 100 men party can kill 300 men. If you have save file I can check.

Example 2 : Currently there is no AI feature of joining siege of an ally army. Means these 164 men will not consider joining your siege as secondary army (can be added later). When Battanian army come closer to them probably they need to run away I can check why they did not if you have save file.

Still we have issue of armies going deep into enemy territory in some cases and we increased importance of distance at AI calculations. Maybe it will help making these cases rarer.

Raids are not that unimportant. If you raid a village it has tons of side effects. Village cannot produce recruits, goods (tax), food for 10-20 days and this effects enemy negatively. Village lose hearth, village notables lose power (which effects recruit production frequency) and also town bounded that village lose loyality and security.
 
Last edited:
mexxico, your dedication to your work and this forum is amazing, considering you are posting in the middle of the night.

But on topic, bad AI is a big reason for me not playing as much as I would like. It is simply too stupid.
Would it be possible to focus on improving AI to be more realistic and having more parameters included in the decision making rather than implementing new features?

Features are great and all, but improved AI is just one of those QOL things that can't be ignored
 
...
I am not sure to which degree it is already supposed to be that way, but armies should be formed
a) to fight enemy armies and
b) to take enemy cities.
and be disbanded when they cannot do either of those things efficiently.
...
Wiser words have never been said. I wonder how easy it is to mod this...

Armies should never be raiding villages. Raiding villages should always be the purview of a single party.
 
mexxico, your dedication to your work and this forum is amazing, considering you are posting in the middle of the night.

But on topic, bad AI is a big reason for me not playing as much as I would like. It is simply too stupid.
Would it be possible to focus on improving AI to be more realistic and having more parameters included in the decision making rather than implementing new features?

Features are great and all, but improved AI is just one of those QOL things that can't be ignored

When you see any bad decision just mail me ([email protected]) your save file with an explanation (something like : here army should defence X rather raid Y) I will try to create time for examining it, sometimes I can examine it 1-2-3 weeks later according to avability, there can be special reasons for AI to do something which seems you unlogical. Other than this I want to spend time on improving campaign AI too but each addition / improvement also costs more miliseconds to us and we need to do all these calculations for hundreds of party at map in so limited time. Actually there are people who want us to change AI to make less calculations than now (even they say AI can randomly can give up evaluation of some targets - which can result in total disaster) because it seems we need to get 30 fps at consoles. We have about 6-8 ms in each frame to do all calculations for some percentage of parties at map (we do not make calculations for all parties each frame we spread them to different frames) and we have two layer AIs for each party on map (long term AI (besiege / raid / defence / patrol / visit settlement / create army for (besiege / raid / defence / patrol) / join army) and short term AI (engage party / avoid party)) and there become so much AI calculations per frame especially when time mode is at fastforward (5x). However improving AI can be done with changing even some constants we are using currently so save files helps me understand what bothers you.
 
Last edited:
If you could set up a debug type system to tell us what the AI is “thinking” it would go a long way towards giving us more of a narrative of world events. As it is everything just feels arbitrary
 
One thing to keep in mind is that the most current update will bring a rebalance for single parties to focus on forming armies for sieging targets over raiding. I do agree that raiding is extremely detrimental to the enemy and is a valid tactic, but I wish there was less of a focus on that and more on finding other parties and sieging targets.
@mexxico: I've also had issues where my factions has an army around the same size as another enemy army nearby and they will pass them by to go siege a city while the enemy army does the same to one of my cities. I didn't know about sending you save files, but how close to the incident does a save file need to be. I will keep this in mind in the future if I see it again.
 
When you see any bad decision just mail me ([email protected]) your save file with an explanation (something like : here army should defence X rather raid Y) I will try to create time for examining it, sometimes I can examine it 1-2-3 weeks later according to avability, there can be special reasons for AI to do something which seems you unlogical. Other than this I want to spend time on improving campaign AI too but each addition / improvement also costs more miliseconds to us and we need to do all these calculations for hundreds of party at map in so limited time. Actually there are people who want us to change AI to make less calculations than now (even they say AI can randomly can give up evaluation of some targets - which can result in total disaster) because it seems we need to get 30 fps at consoles. We have about 6-8 ms in each frame to do all calculations for some percentage of parties at map (we do not make calculations for all parties each frame we spread them to different frames) and we have two layer AIs for each party on map (long term AI (besiege / raid / defence / patrol / visit settlement / create army for (besiege / raid / defence / patrol) / join army) and short term AI (engage party / avoid party)) and there become so much AI calculations per frame especially when time mode is at fastforward (5x). However improving AI can be done with changing even some constants we are using currently so save files helps me understand what bothers you.
Very good and informative answer and this criticism is at no way directed at you, but rather the people making these decisions. You are an example to all game developers mexxico.

BUT (and I'll try to say this in a nice way)

Making AI more stupid (by removing decisions / not having enough parameters) and giving people a worse gaming experience on PC (which is your main community) only to cater to console gamers... do I really have to say out loud that this probably isn't the best decision? You are going to alienate your PC community and the game will suffer for it.

Honestly I would say, just make a product for PC that you can be proud of, and figure out how to port it to consoles later, rather than doing both things at once. It works for Rockstar to make console games and port 1-2 years later, so why wouldn't it work here?
 
#becausemoney
I have to agree with you that this is pretty much always the reason, which seems stupid considering what Callum has previously posted.

Callum:
We shipped over 10 million units of our previous games, which entirely funded the development of this game. We were never at any risk of going broke and could have continued development for years to come without any money troubles.

Finally, some of your math might be a bit off in guessing how much money Bannerlord has brought in so far as you are failing to factor in regional pricing, taxes, etc. However, even when you do factor those in, it is safe to say we are extremely pleased with the hugely successful launch and we are looking forward to supporting the game for many years to come. :smile:

They could have supported development without releasing game to early access for several more years, and now they have even more money due to early access sales which was a hugely successful launch.

Seems to me like they could easily focus on making PC game good now and dealing with an alternate console version of the game later. They SHOULD have the finances to support development of this game in this way. Anything that goes against this is either lazy or greedy. And for anyone chalking it up to "convenience", that's just another word for lazy.

Again, not blaming any developers as it's always the people at the top making these decisions, and the game IS good.
I just wish it could be better.
 
Raids are not that unimportant. If you raid a village it has tons of side effects. Village cannot produce recruits, goods (tax), food for 10-20 days and this effects enemy negatively. Village lose hearth, village notables lose power (which effects recruit production frequency) and also town bounded that village lose loyality and security.
I get that on paper it has an effect, but to me it seems the difference is not enough for the AI. They still have too many recruits and too easy money. They behave as if they have no concern for money whatsoever. Even with zero fiefs they just run all around building parties and even ransoming imprisoned lords. Probably just using their free cheat money for everything. To me the only thing that slows an AI factions down absolutely crushing all their parties. There may in fact be many effects of raids and their certainly are many annoyances for the player's own faction, but for fighting the enemy it does not have noticeable effect IMO. There's always too many AI parties, nothing changes that but executions.
 
I agree that raiding villages are not unimportant, but when small solo parties venture deep into enemy land (why they never stick to the frontline?) to raid, only to get absolutely obliterated by the enemy, that can't be considered good decision making.

Also raiding villages rather than helping armies fight can't be considered good decision making. What is going to determine the outcome of the war more? Raiding a single village or losing your entire army and having no way to make a comeback? :smile:
 
Back
Top Bottom