AI terrible... again.

[Hamster1810]

We all know of how dumb AI be... with all the recently updates.
Especially when it comes to strategic movements... I once brought up this problem, and here I go again for a better view of this.

At the 1st picture, you can see that my forces (The Black Kirko at right corner) is out-numbered. I had only 3 stacks with mostly t1-2 units... While the Orange AI had more than 5 stacks with many mechanic units and a Psionic tank t3... When the AI marched closer to my Colony, the only option I had was stay defensive in the Colony and let they took my sector... until a deadly mistake was made. A t1 unit was left at a single stack under the red arrow while their Commander stay right next to that unit (Green arrow).
All I had to do was surround and destroy that unit together with the Commander stack. That was quite an easy combat. After that, the AI forces was scattered... and with all their tactic points was spent, I managed to destroy them all with the same strategy "Devide and Conquer".

Next picture, the same situation. I was out-numbered at first... then the AI made a terrible strategic movement, they left the red arrow stack exposed to my forces. I again, destroyed together those 2 stacks and left the others vulnerable to be flanked.
After 2 first stacks was easily out-numbered by my 3 stacks, the others was all scatterd. The next target was the Green arrow stack with their commander.

The result.

The problem is, if the AI dont make such mistake, I dont think I can defeat them. My forces in all pictures was mostly t1-2 units, so a direct attack 3-3 stacks will be costly even if I managed to win. The AI just need to keep their forces closed to each others and my only option left is stay defensive.

This happened some times with AI defenders too... 4 stacks to defend their capital, with militia and turrets, one was exposed to my forces, eazy game. (If they had only 3 stacks and stay close, my assault, no doubt, will be more trouble and costly).

 

[iDreadReapr]

i suggest to play vs real players in our multiplayer community: https://discord.gg/7G9A6kt

 

[Tropical Sunset]

I don't think an improved AI is on Triumph's to-do-list.

 

[TibblesTheCat]

I don't think an improved AI is on Triumph's to-do-list.

It is, in the Trex update notes it specifically says so for coming patches.

AI and Diplomacy is an on-going thing we are looking at. We’ve been working hard to improve the AI’s aggression levels and its ability to construct more varied armies with higher tier units. Diplomacy has also been improved with a number of new features, however we ran out of time before we could make a lot of the changes we have planned, so there will be more improvements in future updates!

 

[Fluksen]

I'd be glad if they manage to fix this, but at least it has it's stacks next to each other. I decided in my games against the AI to just ignore this specific type of mistake and pretend the AI hadn't done this stupid move so I'll not take advantage of it instead I'll attack to get a proper 3vs 3 fight (or 3vs4 if you feel like it) in those cases to make it more of a challenge. Otherwise you'll just have to follow Dreads advice and play against human opponents...

 

[Iguanaonastick]

I'm sure Triumph is working on improving the AI's construction patterns and general behaviour, but moving its troops intelligently on the strategic map seems to be the biggest challenge in all games of this sort. Indeed I'd say the Planetfall AI is better than most in that it at least tries to put its troops together in a defensible formation.

From an AI perspective, these things are tricky. If you make the rules too strict the AI might not attack at all, or move too slowly as it keeps its troops in perfect formation. If you make the rules looser, you end up with mistakes like this. If you manage to make the AI fully calculate all permutations so it can move perfectly, AI turns start taking way too long. You have to find some middle ground.

So essentially, I do what Fluksen does and try not to take advantage of the really obvious AI glitches.

Then again, I too sometimes make a mistake when moving a stack and get pounced on by the AI.

 

[Hamster1810]

It's a bit like "a bug"... though I'm not sure to fully understand the difference. The AI, at first, work nomarly.... then for unknown reason, they splited the forces into small frag like this. The 1 unit stack in the picture actually was from 1 full stack.

 

[Ericridge]

I have extensive experience with moving in perfect formations. The biggest problem of this is that it causes the army travel to move very slowly even when they're all of same movement speed on strategic map. A full army group is what, seven stacks right? That's 42 units. I have encountered situations where I had seven full army groups trying to reach the same target but, due to mountains and such Only one army group could pass through and others had to wait.

Other option would be to simply merge all stacks into a single massive stack and establish a unbroken line although that would've came with a disadvantage of breaking up the planned organization tactics of each stack's fighting style. The experience came from aow3 golden realms expansion xD

Judging from the screenshots its definitely the classic case of rough terrain breaking up the AI's planned formation as usual in many turn based strategy games.

The correct answer for the AI is to take more slower and march slowly while maintaining formation the significant drawback of this strategy marching speed is that it gives the opponent time to response and call in reinforcements which is why if the attacker commits to such a strategy for the attack, it has to be done with overwhelming force, with reinforcements continually coming from attacker's territory.

That way it wins even when the defender manages to get some reinforcements in time.

Answer isn't to simply slap roads everywhere because it will cause the AI to make them move its army in a column and be vulnerable to pincers and be destroyed piecemeal as a result.

The sad truth is that to move in correct formation over bad terrain will always slow the army down no matter what but that is the price the attacker pays to ensure that it doesn't get taken advantage of. There is no way around this. Unless you're autonom and everything you have hovers

 

[BloodyBattleBrain]

Thoughts for future games, to deal with these situations:

• Up the amount of units allowed in a stack, while keeping the number of units controlled by players more or less the same (would also result in less clicking in the end game.)
• Expand the adjacent hex rule, for example making it adjacent hex plus one. This is zone of control, as seen in total war games.

 

[Mauer]

IMO, if the game simply dragged all nearby armies to battle (maybe capped at 3 or 4 on each side) instead of only adjacent ones, late game battles would be more even and less exploitable.

 

[Snorro Soliman]

Thoughts for future games, to deal with these situations:

• Up the amount of units allowed in a stack, while keeping the number of units controlled by players more or less the same (would also result in less clicking in the end game.)
• Expand the adjacent hex rule, for example making it adjacent hex plus one. This is zone of control, as seen in total war games.

Or perhaps zone of control as an expensive, tier IV unit ability or Hero skill (think Kwitsatz Haderach foresight, Mentat strategetics, or Creed and his tactical genius). So bringing such a unit allows choosing the combat sphere. Should only work on offensive moves though.

 

[Ericridge]

Thoughts for future games, to deal with these situations:

• Up the amount of units allowed in a stack, while keeping the number of units controlled by players more or less the same (would also result in less clicking in the end game.)
• Expand the adjacent hex rule, for example making it adjacent hex plus one. This is zone of control, as seen in total war games.

I am vehemently against allowing the AI to have 8 units per stack if I can't do the same.

 

[Fluksen]

I am vehemently against allowing the AI to have 8 units per stack if I can't do the same.

Up the amount of units allowed in a stack, while keeping the number of units controlled by players more or less the same (would also result in less clicking in the end game.)

It says more units per stack but same amount of units total, in the end that just means less stacks (and less clicking). This is not suggesting giving bigger stack sizes just to the computer. In case i am the one misunderstanding BBBs suggestion I agree with you though.

 

[BloodyBattleBrain]

It says more units per stack but same amount of units total, in the end that just means less stacks (and less clicking). This is not suggesting giving bigger stack sizes just to the computer. In case i am the one misunderstanding BBBs suggestion I agree with you though.

#

Correct.

I don't know how @Ericridge came to that particular conclusion!

For example, if in a normal game you end up with 60 units under your control, that is 10 stacks.

Upping the stack size would mean those same 60 units are now in 7.5 stacks, which means less clicking for you.

It also means less stacks for the AI, meaning less opportunity for split stacking.

 

[HousePet]

In practise, I don't think that would make any difference.
The AI splits stacks by accident when it tries to use all available movement points (or shuffles units around). Increasing the number of units in a stack is also going to increase the potential number of units that might get left behind by its carelessness.

 

[Ericridge]

#

Correct.

I don't know how @Ericridge came to that particular conclusion!

For example, if in a normal game you end up with 60 units under your control, that is 10 stacks.

Upping the stack size would mean those same 60 units are now in 7.5 stacks, which means less clicking for you.

It also means less stacks for the AI, meaning less opportunity for split stacking.

I came to that conclusion because you said up the number of units in a stack while keeping player's stacks at more or less the same which means no change to me which sounded like players had to keep their stacks at 6 units per stack while AIs was allowed to have more. That's why. It wasn't spoken very well in my opinion.


I think the easiest way to deal with stack splitting is because human players have no choice in the matter. 18 units is going to have a rough time fighting against 48+ units at same time and AI's blunders in manual combat is what saves the players. From what I'm seeing against the extreme AI, it seems that players have to for example pay 100 production for a single trooper but the AI only have to pay 20 production for same trooper. Thus Overflow mechanics happen very often for the AI. For me, it doesn't really happen until a city is really developed.

I got derailed a little bit there, anyways.

Best counter to the stacksplitting is to simply drag in all stacks in a certain radius around the point of contact. This means battlefields would need to be bigger. As a guy who used to fight with hundreds of units on small battlefields of aow3. It would be unbearably small xD

And some of the maps in planetfall would be very scary when fighting a massive battle like that lol Heavy weapons factory on the surface not dungeon map version one for starters. Everyone basically starts on the high ground in there and then in the middle of map is the low ground. Units with many numbers of indirect attacks is quite powerful in there. Once fought the growth in it.

Everything is currently balanced around the fact that there is max of 8 stacks + militia + operation/anything else as reinforcements.

Another thing, every unit being pulled in from a certain radius means that we would easily see 8 stacks vs 8 stacks battles. At very least, the size of battlefields would need to double at minimum. And once the numbers get that high, operations would need to be rebalanced because they would be silly tiny for rather big battles. And the turrets in cities would need to be rebalanced as well.

I'm having fun imaging what the battles would look like.

 

[Sandman25DCSSS]

I wouldn't like increasing stack size, at least as long as adjacent hex rule is in place. The game would be much better for me without those huge 18 vs 24 combats.

 

[BloodyBattleBrain]

I came to that conclusion because you said up the number of units in a stack while keeping player's stacks at more or less the same which means no change to me which sounded like players had to keep their stacks at 6 units per stack while AIs was allowed to have more. That's why. It wasn't spoken very well in my opinion.

Up the amount of units allowed in a stack, while keeping the number of units controlled by players more or less the same

Sorry you got confused. To clarify, when I say "number of units controlled by players more or less the same" I am referring to how many units you have in your empire, which is limited by economy and recruitment sites. For example, in a typical game of Planetfall (I play large or extra large maps, max enemies) I will have maybe 40 - 50 units total towards the end game, usually in sets of 3 or 4 stacks (18-24 units in an area.)

That's a fair amount of clicking, which I think could be reduced by having fewer stacks. The logical end point is something like Total War, but imho that formula has it's own issues, and gets quite tedious moving your mono stack around...

The AI splits stacks by accident when it tries to use all available movement points (or shuffles units around). Increasing the number of units in a stack is also going to increase the potential number of units that might get left behind by its carelessness.

That's a good point. However, I don't think this is really fixable, short of redoing the game from the base code up? Which is where this comes in:

Expand the adjacent hex rule, for example making it adjacent hex plus one. This is zone of control, as seen in total war games.

The intended result here would be to make it rarer for the AI to leave dangling units, and harder (zone of control) for a player to take advantage of it, and more dangerous (because you won't fight dangling unit+ 1 stack, but more likely dangling unit + the other stacks.

I wouldn't like increasing stack size, at least as long as adjacent hex rule is in place. The game would be much better for me without those huge 18 vs 24 combats.

These large fights are meant, as far as I know, to be towards the end of the game, the culmination of your military campaign, the deciding battle. As in you're not meant to have too many of them. Which is why the late game in AoW3, with larger maps especially, started to break down, because you'd get to the point where you'd have a large battle every turn, sometimes multiple in a turn. And having multiple t4 units doing this was also there were so many complaints about "t4 spam" (the Ai would put 2 or so t4 units in a stack, but when the AI has 12 stacks, it could certainly feel like "spam")


Planetfall handles this much better, and the battlefields are larger. So, imho, there is room for larger battles.

My idea is that there are fewer battles towards the end game, but each battle is more significant.

Also, when this is discussed, people seem to frequently forget that 8 units per stack was the standard. AoW3 changed that, but the original game had large maps, very restricted t4 and t3 recruitment, 8 unit stacks and split stacking the AI was, by memory (ergo fallible, ofcourse) was harder then. It was much more challenging assembling a large army.

Sadly, t3 and t4 units were crazy powerful (too powerful imho) and heroes and leaders were ludicrous.

A better blend for a future game would be to use Planetfall as a foundation (it is a crazy good game imho. Too many great improvements for me to bother listing) and further tweak the interplay between t1,2,3,4 units, and potentially introduce T5, T6 units (legendary, monstrous units ) but address the biggest issues (again, imo) with the game, which are, to be fair, very hard problems, namely:

• AI idiosyncrasies
• Simultaneous combats

 

[BloodyBattleBrain]

Everything is currently balanced around the fact that there is max of 8 stacks + militia + operation/anything else as reinforcements.

This is a very good point. Game would have to be retooled a bit.

Best counter to the stacksplitting is to simply drag in all stacks in a certain radius around the point of contact.

Expanded adjacent hex rule.

I expanded on this in a post long ago, but attach a delay to armies outside the adjacent hex to allow things like a raid.

For example, I attack your army. The attacked army and the adjacent one start the fight (current system) and 2 turns later, the army next to that shows up. And 2 turns later, the next army.


In 4 turns, the attacking army could do alot of damage if they wanted, let of a bunch of spells. It would, imho, actually add in another layer of thinking.

And some of the maps in planetfall would be very scary when fighting a massive battle like that lol Heavy weapons factory on the surface not dungeon map version one for starters. Everyone basically starts on the high ground in there and then in the middle of map is the low ground. Units with many numbers of indirect attacks is quite powerful in there. Once fought the growth in it.

Another good point. Triumph have already changed how dungeons work, so for them, I think this is fairly easy.

 

[Ericridge]

Expanded adjacent hex rule.

I expanded on this in a post long ago, but attach a delay to armies outside the adjacent hex to allow things like a raid.

For example, I attack your army. The attacked army and the adjacent one start the fight (current system) and 2 turns later, the army next to that shows up. And 2 turns later, the next army.


In 4 turns, the attacking army could do alot of damage if they wanted, let of a bunch of spells. It would, imho, actually add in another layer of thinking.

That kinda won't work very well. Let me tell you how. Say, you are the defender and your dangling stack got attacked? Best course of action? Is to immediately flee to the nearest friendly retreat side of the battlefield and wait for your allies to show up and refuse engagement. This would be very easy to achieve. There is zero risk in this, simply because there's no very long ranged shots in planetfall like 20+ range. And if such a thing shows up, then superior tactic would be to whittle down the defenders with long ranged shots while maintaining range.

A raid would only work if combat is forced.


Total War approach is not really 100% ideal for planetfall. The game is 2002 edition of medieval total war. This is very old ancient story of mine that I'm about to say but it achieved insane victory. I was playing the byzantines aka eastern romans and the golden horde khanate showed up. On very same turn I ferried every soldier I had in my empire to meet their horde of 200K+ with only 20k soldiers at my command at max. Battle begins, the map drawn had a river in the middle with many mountains and a couple of bridges plus a forest on khan's side. Starting army on both sides is 1k byzantines vs 1k golden khanates.

I send 500 sword shield heavy infantry + 200 heavy cavalry / archer to guard the bridge in the middle of the map. AI send all 1000 of their troops to try take this bridge from me and kill me. I send 300 heavy cavalry on a long route around a forest and crossed the second bridge completely undetected and then snuck through the khan's forest while my 700 was just buying time and keeping the AI's attention focused on me.

After 300 heavy cavalry worked their way through khan's forest, they felled upon the golden khanate's 1k. Swift rout occured and I took the khan king prisoner plus all his sons. I chase the fleetest footed remnants of his army all the way to his reinforcement point. Then Both of our reinforcements finally start to come into play. The shock of seeing my byzantine troops upon khanate's reinforcement point just routed 200k+ solders instantly. 200k+ prisoners captured. I press execute prisoners button and I acquire a bloodthirsty reputation on my emperor. The ransom for that many soldiers + nobility + damn king and his sons all at same time would've been ludicrous. I took the I won button instead.

In total war, soldiers if they see friendly soldiers fleeing the battle, they get morale debuff and would want to flee. This would transfer to like -600 morale penalty to reinforcements who is trying to support the army stack that got raided in planetfall.

On world map, the golden khanate got destroyed on the same turn as they showed up. Fate of empires decided on the actions of very few.



That is the story and that is why I don't think the raids would work very well when translated into the games. Because when one is raided, the correct action is to take defensive posture and get support to rout the raids. Which in turn makes attempting to launch raids rather meaninglessly, at most only a position on battlefield is achieved. In planetfall, this would basically mean trying to force the raided party out of what position they are on battlefield and you take it from them and defend it because it gives you fantastic firing lanes, etc needed supremacy to help you defeat the rest of the attackers.

I think a good example of this would be that one place on arid map where the middle of the map is mostly neat and clean with excellent firing lanes while all the positions around the middle is terrible and has tons of obstacles that prevent an opposing army from bringing a proper gunline to maximize dps in a single turn.