Suicide commanders - a game killer?

[Scorpion]

After getting the game on the day of its release, I played it a bit and noticed a familiar trend from Total War releases. The AI fearlessly throws its army commander into the fray at all the wrong places and times, generally getting slaughtered and causing his army to rout.

I´ve heard the same happening if the AI commands your reinforcements, causing the death of many a promising character.

Now, the AI isn´t really giving the player a truly hard time tactics-wise as it is, but compounded by this matter battles seem to be a bit easy to win (I play on Hard, as I don´t believe in giving unfair stat bonuses to the AI soldiers)

I haven´t played much after these initial impressions, because I´m positive that CA will fix this and several other issues in the patch, but the question is - when?

I´ve been asking myself whether I should still play, or wait until the patch. I´ve got plenty of other things to do, but still...

What do you think? Have you found this issue to be annoying, and might there be any solutions before they fix it? Perhaps modding?

[Steppe Merc]

I think it's better than the uber Jedis of Medival. And it hasn't been that much of a problem for me, but then I use horse archer armies so it's hard for my enemys to actually commit in a fight against me.

[Red Harvest]

Considering how frequently this problem has happened in previous TW releases (just about every one). I'm not sure that waiting for a patch would make much difference. I really have to wonder what about the AI programming prevents it from trying to preserve the commander? It uses this heavy unit as if it were just the heaviest unit out there, and smashes away with it. Every castle siege the commander meets me at the entrance point (unless there are so few units that he is stuck guarding the plaza.) Come on, this is just dumb and sloppy. In fact, it is not even smart about this employment even for a heavy non-command unit.

What the AI should be doing is some sort of determination before the battle like this:

1. Can I afford to lose this commander in the battle?
2. Are the odds favorable, even, unfavorable?
3. Can I retreat without losing him getting him trapped to where he can't be helped, etc? (Heroic last stand.)

In battle, the AI should be committing the commander only when it can afford to, and when it has done some sort of evaluation to conclude committing him will decide the battle in the AI's favor (or all is lost and it can cause some real damage going down swinging.)

[Morindin]

I avoid at all costs of having two of my armies near each other with two generals. If I am going to have a supporting army I remove the general from it and send him somewhere else, otherwise he will get killed and my "reinforcement" army routs.

Now while having AI reinforcements removes the problem of having large armys and only 20 slots, and makes it somewhat more "realistic", I much prefer the old MTW system.

[Basileus]

Alot of ppl have had problems with this it seems though it has never happend to me.

[Jeanne d'arc]

I had reinforcements once when i was besieged by the gauls, the general's army joined the ranks of my army and waited untill i attacked the gauls.
I also had a few normal battles on the field where my reinforcements general actually saved the day for me, he did join in the fighting but thats to be expected from good generals.
No suicidal generals so far.

[Haido]

I think this is a common thing to happen to less experienced AI commanders.

I've fought lots of AI commanders that wait and lead their troops.
I've fought many who withdrew all their forces just by seeing that my force was more powerful although not in troopsize and so on.

I play julii campaign and usually when i fight enemy AI it's just some captain or lowlevel family member. But the times ive met 5 star and more they have often been really good.

[Morindin]

I think this is a common thing to happen to less experienced AI commanders.

I've fought lots of AI commanders that wait and lead their troops.
I've fought many who withdrew all their forces just by seeing that my force was more powerful although not in troopsize and so on.

I play julii campaign and usually when i fight enemy AI it's just some captain or lowlevel family member. But the times ive met 5 star and more they have often been really good.

Actually come to think of it, this may be the case. Ive had the AI withdraw on me plenty of times too.
However when determining the strength of an army im sure that slider doesnt take into consideration how many stars each general has.

90% of the time though you're fighting AI captians or a few star family members, out of over 200 battles in my campaign so far I can count the ones on my hand that have been against a general with ANY stars.

[Tamur]

I've run into the charging general a few times out of a bit over 100 battles. Most of the time they either hang back, or try to flank with their speed (when they're horse, that is).

I'd also say it's a matter of star rating. The main problem is that the AI armies are most often led by commanders with zero stars... apparently they don't get into battles quite as much as we players do. I do remember fighting a 7-star Gaul general, and it was a very tough fight (on Hard) even though I had him outnumbered 3-to-2.

[Hamburglar]

The problem seems worse than in MTW.

The Jedi Generals in MTW at least had the side effect of suicidal generals living a bit longer than normal.

Now in RTW the generals are just as suicidal except they die so much quicker.

I got really dissappointed in my Julli campaign. I finally started the civil war and got in a HUUUUUUUUGE battle with the Senate. Senate Faction Leader goes up BY HIMSELF and gets utterly massacred by my archers before his infantry could even catch up to join the fray.

[Red Harvest]

I have not kept stats, but from what I've seen, if the general did not withdraw, then about 50% of them have done something suicidal like a banzai charge. They like to chase my slingers, or cav that I have used to kill off skirmishers, archers or end units. They don't seem to have any respect for my masses of cav...Roman generals are horrible about this.

[Colovion]

I think that it would be a revolutionary shift in AI of any sort if they were able to realize the danger present by weighing their stats up against their opponent they are about to charge - factor in outside influences such as archers present/in range and other units surrounding them and their disposition (in combat, standing, fresh, shaken, out of ammo etc..) and on the basis of the results the computer would do whatever their General's traits dictate and what a smart commander would do with their troops... or something

You'd really think with these amazingly fast computers we have these days that the AI would catch up to it.

We can only hope that CA is just working like busy little bees on some huge patch to fix the AI which is only marginally improved from MTW. Granted - I know nothing of programming so I don't know how difficult it would be which would generate such a believable AI.

[Scorpion]

Hm. The comments so far seem pretty depressing about the matter.

Perhaps it´s just best to sit and wait it out - it´s not like there aren´t many other things the patch needs to fix.


I guess this is what we should have come to expect from CA - brilliant games, but flawed by buggy releases (and somehow all the reviewers seem to ignore these matters, but I guess it is for the best. Wouldn´t want to discourage people...)

[Jambo]

I wouldn't exactly call RTW flawed and buggy. Compared to other strategy games of a similar ilk, it's much more stable and polished. Consider that EU2 has had somewhere around 5-6 patches and Civ III more, and for Civ III this was just to make it playable!

[RJV]

I wouldn't exactly call RTW flawed and buggy. Compared to other strategy games of a similar ilk, it's much more stable and polished. Consider that EU2 has had somewhere around 5-6 patches and Civ III more, and for Civ III this was just to make it playable!

Problem is that patches from CA are not exactly earth-shattering in their speed of release. We play for a long time with the 'buggy' version (though to be honest most of the bugs I can live with, I'm fairly easy going) before the fix comes out. Now whether you regard the bugs in the game at the moment as critical or not, past experience suggests that it's going to be a while before they are fixed. Admittedly this is a far better situation than the ChampMan4 where there was a patch ready the day BEFORE the game was released, then another one straight away (which gives you some idea of the number of errors in the thing), but I'd like there to be a happy medium. Initial play has revealed a number of issues, and I REFUSE to believe that the testing process at CA/Activision didn't throw up any of these thus making these ALL 'new' problems. Rambling now, will stop.

Cheers,

Rob.

[Longasc]

There are many AI flaws in the battles:

1.) Generals die too quickly.
Their troops do not protect them, they do not care much for their lives and die

- this usually sents the rest of the army routing, especially weaker enemies like the Gauls

2.) The AI waits for you to encircle it completely. Even MTW's AI walked simply higher or attacked or did something, in Rome it just waits.

3.) Sometimes Units just stand around and wait to die of arrow fire - till death.

4.) CTRL+A and rush through the gate/breach and you will probably win the Siege even if outnumbered.

5.) Cavalry Charges are powerful without end. I did not try Elephants yet, but I fear:

Battles usually end up with the "loser" having a loss ratio of 99-100%.

(Hard/Hard)




Add to this that I can bribe a full Army of Principes and Early Legionaries for some 3000 Gold with an untrained Diplomat.


This is really sad, as these factors all make the game worse than MTW, while it shines in so many aspects.

Dumbed down and sped up battles plus some flaws of the overall really greatly improved strategic map / interface (Micromanagement difficult due to lack of "cycle spies" button e.g.) are just sad.


Be honest - if this is 92% or 109% of 100... then there is something seriously wrong with the Hype!

[Jambo]

maybe you should try Very Hard/Very Hard ?

[Ii Naomasa]

Surprisingly enough, while a lot of enemy AI generals will die because I make an effort to target them if it won't detract from the battle, my biggest issues with suicidal generals have been on my side. Old/young, new/experienced, no star/five star, offense/defense, it doesn't matter. They all seem to think reinforcing means charging out of the gate at top speed (meaning it's always the cavalry (including the general) who hits the enemy first and gets slaughtered). Even enemy AI reinforcements seem to be a little more cautious.

As someone else said, whenever I have two general-leading armies near each other, I either drop one general or move them to the same group in order to avoid worrying that one will just decide it's time to die. It's pretty sad, actually.

Of course, having a captain-lead group too close to a general group usually means you have the captain being attacked and the general under AI anyway...it pays to always be on the offense, I suppose...

While Roman history has a lot of examples of generals who were so opposed in how they fought that they didn't really work as teams, it would be nice if there was an option for you to give the AI at least an overall strategy, such as telling them to be aggressive, be defensive, or hold back. I once had a beautiful hillside defensive position and was tearing the enemy with archers and onagers and suddenly my buddy rides through my lines (taking casualties, of course) and charges into their lines. At that point, I was stuck with the choice of watching the AI throw away my secondary force and its good commander, or have my men also take a lot more casualties than they should and charge in for the rescue. That's not what reinforcements are for...

[The_Emperor]

The only instance of suicide general I have found was when a Mecedonian General frontally charged my Hoplite Phalanx in some tight city streets, but he deserved it!

I once had a battle in my Julii campaign where my AI reinforcement army was larger than the one I was in command of. My other general started off on the side of a hill, I started off right at the bottom of it... Where was the enemy? Thats right at the top of the hill.

My reinforcements reached them first and their lines clashed, while my men were still slogging their way uphill. I was impressed to see that my general did not charge in right away but sent in the Equites first on the higher ground and outflanked the enemy...

By this time my army finally came up the hill and engaged on the other flank and the Gauls ran away Screaming like little girls. The Hounds were released, and the foxhunt-like scene began.

I was very impressed with how my other army fought in that battle, I almost didn't get a look in til the end!

[Spino]

Rome is a fantastic game but it's things like this that will adversely affect its long term playability unless fixed. Modders can change many things but the AI is not one of them. AI controlled general don't need to be tougher, they simply need to get rid of their death wish.

Last night I set up a custom battle to test out some tweaks I made to the elephant stats. My Julii army was on the defensive versus the Selecids. Out of curiousity I placed my army in the AI's control to see what would happen and boom! While the bulk of the army assembled in a defensive formation my general and another cavalry unit charged directly towards the Seleucid's bristling phalanx wall... even with Adonys' half speed kill rate mod the General's unit lasted all of 10 seconds. How did this get past beta testing?

maybe you should try Very Hard/Very Hard ?

If Rome's difficulty settings are anything like Medieval's then changing it from Hard to Very Hard shouldn't improve or change the AI but will simply give a large morale boost to its troops in battle.

[Longasc]

The thing is, how could the AI of Rome become worse than that of MTW?

It is probably a completely new AI, as its behaviour is very different, and unfortunately it is worse.

[Scorpion]

maybe you should try Very Hard/Very Hard ?

The problem is, as far as I understand, upping to very hard does not mean a better AI opponent, it just means the AI cheats more.

And I´ve never liked that....

[Morindin]

There are many AI flaws in the battles:

1.) Generals die too quickly.
Their troops do not protect them, they do not care much for their lives and die

- this usually sents the rest of the army routing, especially weaker enemies like the Gauls

2.) The AI waits for you to encircle it completely. Even MTW's AI walked simply higher or attacked or did something, in Rome it just waits.

3.) Sometimes Units just stand around and wait to die of arrow fire - till death.

4.) CTRL+A and rush through the gate/breach and you will probably win the Siege even if outnumbered.

5.) Cavalry Charges are powerful without end. I did not try Elephants yet, but I fear:

Battles usually end up with the "loser" having a loss ratio of 99-100%.

(Hard/Hard)

Add to this that I can bribe a full Army of Principes and Early Legionaries for some 3000 Gold with an untrained Diplomat.


This is really sad, as these factors all make the game worse than MTW, while it shines in so many aspects.

Dumbed down and sped up battles plus some flaws of the overall really greatly improved strategic map / interface (Micromanagement difficult due to lack of "cycle spies" button e.g.) are just sad.


Be honest - if this is 92% or 109% of 100... then there is something seriously wrong with the Hype!

1) Yes generals die quickly, but MOST generals are captians. The rest of your point only applies to the Gauls.
Most other stronger factions units fight for a looong time with or without their general. I had whittled a Carthagian Iberian infantry unit down to one man last night before it routed.

2) Every time I try to flank the AI it will break some of its army up to counter flank, this implies coming from behind, anywhere. The only times it seems to not do anything is if its defending on high ground or severly outnumbered. With larger numbers, ive seen plenty of times where the AI in RTW has done exactly what its done in MTW, positions itself on higher ground, move its forces out of missile range, send out fast units to engage my flanking cavalry, etc. Ive also seen it do, just like MTW, stupid things. Over all though this part seems improved apart from a few circumstances. See Below

3) Yes, this only happens though if you vastly outnumber it and the AI doesnt do anything, because any option is suicide! If the AI has a large army it wont sit there and let you pepper it with arrows.

4) The AI comes off its walls when you've breached the castle defenses and captured the gate in a siege. Again maybe against Gaul towns this tactic works, against a Carthagian/Greek army holed up with archers in a proper city, doubt it.

5) Cavalry charges get minced by their counter unit from all my experiance, both multiplayer and single player. Charging an infantry line is going to be devistating, but I felt thundering Knights charging an infantry line in MTW and killing maybe 5-6 guys was a bit underpowered. Remember all the cavalry is underpowered threads in the MTW multiplayer forums? Go back and read them.
Cavalry in RTW is undoubtadly much weaker in melee however. Perhaps their charge bonus is too strong for their cost, but the fact cavalry sucks in melee much worse than MTW cavalry ever did is a balancing act and realistic too (no stirrups etc)
War elephants are very strong, but they also cost 2600 and you get 12 for that cost.

6) The ratio in battles in RTW is pretty much the same as MTW 1.0 and MTW 1.1. The only difference is you take no prisoners. Since I hardly took prisoners in MTW, since you were always overflowing with money, the dread vices helped, and it saved you fighting the army again, this is hardly a problem.
Running cavalry though routers MTW 1.1 is the same as running cavalry though routers RTW, except the RTW routers dont magically dissappear like they've been abducted by aliens.
Also the maps are larger in RTW giving you more time to chase down all the routers and win decisively.

7)

Add to this that I can bribe a full Army of Principes and Early Legionaries for some 3000 Gold with an untrained Diplomat.

This is truely an exaggeration or a one off pot luck scenario. By the same account, its cost me over 3000 to bribe one army of those Spanish javelin thrower guys.

[Sparky]

I agree this is still ( ) a problem. Mainly manifests itself in sieges (either attacking or having your AI general join the fray). Sieges are much too easy primarily because the enemy general will throw himself at the front lines, leading to his quick demise and a mop-up operation. In field battles I've seen some pretty good general behaviour, must be something to do with the formations I think. I really hope they look at this, sieges should and would be a climatic struggle if the enemy general would just keep out of the action! I'd rather they'd just hack it so he never engaged (if he could possibly avoid it) while trying to keep fairly close to the fighting, than the current situation.

It's not a game stopper for me, but it is the one disappointment I've had.

[Tamur]

So, anyone up for making a knowledge base of tactical situations, a definition set to fit it, and a really fast algorithm to create new KB rules on the fly? Sounds like a good degree project... hmmm....

[Spino]

For whatever it's worth in custom battles the first unit you select becomes your general. If you select an infantry unit or missile cavalry unit to be the AI's general it either doesn't embark on a banzai charge into your lines or it's skirmish mode won't allow it to.

I suppose if you really got sick of watching the AI enemy and reinforcement generals impale themselves on sharp pointy sticks you could mod the unit stat files and turn the AI general units into javelin or bow wielding cavalry with lots of ammo to keep them out of trouble. Not a neat or historical solution but it should keep them alive much longer and will help the AI's overall war effort until CA corrects this behavior in a patch.

Just a suggestion.