AI behaviour-lines of battle

[GaiusMarius]

Just to raise a point regarding AI behaviour in battles with substantial differencies between lenghts of the opposing lines (Marathon is just an example: this happens anyway regularry in any scenario I've tried, even ones involving heavy infantry or shock trops). What happens is as you can see in the picture attached: AI attacks as expected with all the units who have an Medium or heavy enemy unit in front of them: units without an enemy or only with LF/LH in front of them start to point uselessly to the end of the map (enemy side) without even attempting to participate to battle. I've extensively tested this behaviour by modding leader aggressiveness without any appreciabile result. Needless to say that this reduces a lot the strenght of AI as opponent.
Smaller point you can also see in picture below:makes any sense to have baggage trains attacking as regular units?

Thoughts/feedbaks from anybody are really welcome.

[arsan]

The AI is in need of A LOT of work.
Its one of the weakest Ai i've ever found... and one has a lot of weak Ai's around to choose from
When RoR was initially delayed Slitherine/Hexwar promised it will have an improved AI when it was finally released.
But it seems the promise was forgotten... SoA don't seem to add any AI improvements either

[GaiusMarius]

Thanks for the feedback

Yes, you're right with SoA (and previously RoR) no visible improvements, with the exception of a clear bug involving AI movement of MF with bows solved satisfactorily.

The point you raised is perfectly fitting with my opinion which is: after months passed solving game and rule bugs (clearly priority 1: regarding this I should say only thanks for the good job done from Slitherine/FOG Teams. A good system was developed, bugs were eliminated quickly, nearly always with an eye to the player feedback), now should be the time to take almost some small steps in the direction of a more performing AI. Please note that I'm not referring to radical changes in AI logic, I'm talking about only to near-bug behaviours which, almost for a player like me playing extensively against AI, are really annoying/frustrating.

Kind regards

[lpgamble]

We could suggest some improvements like:

1) don't use dancing bows when only light foot is in range.
2) Always attack the unit with the lowest order. It seems to prefer attacking generals
to any other unit and to often picks a much harder fight than the one it should.
3) Never spend more than 25% on lights if possible
4) After main lines are in range to contact send the 50% lights around the flanks

[TheGrayMouser]

The AI is in need of A LOT of work.
Its one of the weakest Ai i've ever found... and one has a lot of weak Ai's around to choose from
When RoR was initially delayed Slitherine/Hexwar promised it will have an improved AI when it was finally released.
But it seems the promise was forgotten... SoA don't seem to add any AI improvements either

While i agree the AI is no genius, can you please direct me to any turn based, igougo, hex wargame, that is in a timeframe where individual units need to be used in conjunction with their relative proximity to other units for best effect, that has a superior AI? ie any game set between ancients and Us Civil war

If you cannot, then i think your statement is a little harsh...

[Xiggy]

I agree with TheGreyMouser about the above statement being a bit harsh. as far as targetiing priorities, I liek what was said. I think their still should be some randomness to what the AI does. That was you can't predict exactly what it does. You can also handicap yourself, when playing the AI.

[arsan]

The AI is in need of A LOT of work.
Its one of the weakest Ai i've ever found... and one has a lot of weak Ai's around to choose from
When RoR was initially delayed Slitherine/Hexwar promised it will have an improved AI when it was finally released.
But it seems the promise was forgotten... SoA don't seem to add any AI improvements either

While i agree the AI is no genius, can you please direct me to any turn based, igougo, hex wargame, that is in a timeframe where individual units need to be used in conjunction with their relative proximity to other units for best effect, that has a superior AI? ie any game set between ancients and Us Civil war

If you cannot, then i think your statement is a little harsh...

To make comparisons with a game extremely similar to FoG (same era, same scale, same tabletop game translation style etc) Tin Soldiers Caesar or Alexander for example had a very respectable AI for example. Much, much better than FoG.

FoG of course have a great PBEM system and the DAG, but lets be honest, its Ai is terrible.
Just look at the first post screen: assaulting baggage trains, half the Persian army just passed by the enemy line and keep on going without noticing they are there and triying some rear attacks...

Something as basic as "always resolve combats from easiest to hardest" so you can cause a moral check on an enemy BG after routing and adjacent BG will be a help. It don't need any long term planning. Just calculate the odds, and start resolving the easies first...

Another easy tweak that could help to hide AI's weaknesses (even if it will not make it better) on DAG games would be to set for the AI a reasonable basic deployment (HF on the center, cav on the sides, LF in front, keep everybody together forming a basic battle line...).
Nothing fancy, but surely much better than the current three mixed groups without any meaningful order than will just blindly advance over any terrain, no matter how bad for them.

And for the DAG AI army picks... they seem complete random nonsense... Maybe some fixed templates for each army (or army style) could be done for the AI so they resemble something reasonable.

Cheers

[TheGrayMouser]

Hey Arsan, cant comment on the Tin Soldiers game as I never picked them up (specifically because they included no scenario editor at all in their games)

I agree some tweaks to the Ai routine would be great and that, just because noone has come up with a great AI in this genre doesnt mean that no effort should be put into it...

I do try to make my Ai games more challenging especially in DAg battles
A allow the Ai several turns to shake itself out and actually form up a plan instead of beelining right at it
B try to play w the concept of moving and attacking with formations instead of attempting to maximise the efficiency of every single line up for your individual units (human is alway 10x better at doing this)
C dont deploy your forces, accept how they are scattered about, for after all, the AI's are and why do you need an even greater advantage?

Cheers!

[arsan]

Hey Arsan, cant comment on the Tin Soldiers game as I never picked them up (specifically because they included no scenario editor at all in their games)

Yes, indeed they only had some canned scenarios and campaigns. Not as much replayability as FoG, but speakings about AI, they were much better and single player was challenging (in fact, it was too much challenging in some scenarios on which you was outnumbered of had some thigh objectives to fulfill).

The best thing i can say about FoG AI is than hopefully you don't have to bear with it as the PBEM system is great

But having a decent AI woudl be great to test new armies or tactics. But currently is not of much use and what you get has very little in common with a PBEM battle.

Cheers

[batesmotel]

AS a minor observation, I think the AI does seem to do better with higher point values, e.g. 650, than with lower ones, e.g. 400. With more BGs on the field it ends up with something closer to a line despite its best efforts not to.

Chris

[TheGrayMouser]

One thing i might suggest is not to assume an AI is good just because the game is challenging... The AI having overwhelming #'s, hard to reach victory hexes etc all can make a game hard without the AI being any good at all. Even the GMT pc versions of Alexender Hannibal etc were challenging, but the AI was horrible.. Some of those scenarios were like 4 turns long and both battle lines literally were 2 hexes away from eachother at the start! FOG has none of the above mechanics in order to make a scenario harder, just the AI by its lonesome.. I think it does rather well considering....
I too wish the AI in dag battles didnt cluster into 3 brigades and bought a better troop mix

[arsan]

One thing i might suggest is not to assume an AI is good just because the game is challenging... The AI having overwhelming #'s, hard to reach victory hexes etc all can make a game hard without the AI being any good at all. Even the GMT pc versions of Alexender Hannibal etc were challenging, but the AI was horrible.. Some of those scenarios were like 4 turns long and both battle lines literally were 2 hexes away from eachother at the start! FOG has none of the above mechanics in order to make a scenario harder, just the AI by its lonesome.. I think it does rather well considering....
I too wish the AI in dag battles didnt cluster into 3 brigades and bought a better troop mix

Yes, i know. Part of the challenge was in the scenarios setup. But the AI was not a pushover at all.
There was a demo of Alexander aroudn in case anyone wants to give it a try.
I mainly played Caesar, (had better graphics and a non linear campaign), but the battle system was more or less the same in both.

[deeter]

I never played the Tin Soldier games, but looked at them and believe the scenarios were scripted given their static nature. Scripted behavior can yield some challenging games, but scripted behavior is not the same as AI. And scripted behavior falls down in random battles such as DAG-generated games.

Deeter

[arsan]

I never played the Tin Soldier games, but looked at them and believe the scenarios were scripted given their static nature. Scripted behavior can yield some challenging games, but scripted behavior is not the same as AI. And scripted behavior falls down in random battles such as DAG-generated games.

Deeter

No, i think there was no scripts, but on Caesar you had to minimize your losses (as forces were carried from battle to battle, adding some reinforcements) and take some objectives.
But i've been playing wargames and PC games since 25 years ago. I can tell apart what's done by the scenario setup and whats done by the AI: keeping a cohesive battle line, attacking on your weak spots, trying to get to your flank and rear...

[GaiusMarius]

Yes must admit that Thin Soldiers: Caesar is a good game in terms of vs AI playability. Anyway, please don't misunderstand me: I've seen worst things compared to FOG AI (the old Battlegroud series in example). It's also clear that the real strenght of this game is the multiplayer feature. there are anyway unexplainable things especially in the AI movement logic (another example: have you tried to put in a scenario some of your cavalry units not on the same line with respect of the rest of your battlegroups, let's say 7-8 hexes closer to the AI controlled side? well, if you do it AI literally loses his wings, not being able to move them in any useful way. the battlegroups on the center are unaffected by this. I found this in creating a revised version of the battle of Clastidium, before to give up and return to the set up "wanted" by AI to work cohoperatively) which make me think we're facing also bugs and not only flaws in the AI logic. If this is the case (bugs), is a quick win. the AI performance can be improved without the bigger effort needed to revise its logic.

Cheers

[Sull51]

Why not allow the player the ability to switch sides at the begining to overcome the weakess of AI setup and initial play.

[GaiusMarius]

Good idea in order to solve problem one (setup): for problem 2 (first turns "confusion"), I'm not convinced that this is sufficient: unfortunately AI is a problem through all the duration of a game.

Cheers

[Brigz]

I prefer not to play solitaire against the AI but instead play solitaire Human vs. Human. The AI has very little tactical ability and I am very rarely surprised by something the AI does. Allowing me to make moves for both sides at least gives a simblance of historical lines of battle and tactics. The only problem is that it chalks up a loss to me no matter which side wins.