How did the suicidal Generals get past the testers?

[Owen]

|
|
|
|
|
V

Criticising a game for poor ai gameplay is entirely different to criticising a game for pedantic historical accuracy.

[Medieval Assassin]

Criticising a game for poor ai gameplay is entirely different to criticising a game for pedantic historical accuracy.

It's still hurting my baby. . .

[Parmenio]

Ah... ...those suicide generals. I suppose I haven't noted them particularly simply because the enemy AI overall is equally short sighted.

I seen the AI form reasonable battle lines. I've seen it deploy flankers fairly well. I've even see it use skirmishers correctly on the odd occassion.

But ever time it goes for the big melee crunch it just loses it. I've seen solid lines of phalanx suddenly charge apart in different directions losing all cohesion. It'll recklessly charge any cavalry (not just generals) after skirmishers regardless of how many heavy units are packed together right behind them.

It's not broken as such, but the AI could use some more tweaking in general, rather than just Generals.

[Spino]

Yes, suicidal general units are just one of those things TW fans have endured throughout the series. Even when we thought it was addressed via a patch for MTW it comes back again!!! The TW fans have now reached the point where we collectively roll our eyes and groan, "Oh, not this again!"

I seen the AI form reasonable battle lines. I've seen it deploy flankers fairly well. I've even see it use skirmishers correctly on the odd occassion.

But ever time it goes for the big melee crunch it just loses it. I've seen solid lines of phalanx suddenly charge apart in different directions losing all cohesion. It'll recklessly charge any cavalry (not just generals) after skirmishers regardless of how many heavy units are packed together right behind them.

Exactly. The more I play RTW the more disappointed I am with CA because they didn't overhaul and revolutionize the tactical AI in RTW in the same fashion as strategic map and the graphic engine. I was desperately hoping that RTW's AI would raise the bar considerably, with the tactical AI setting a new standard in the industry. And yet the reality is that we have a tactical AI opponent in Rome that is in many ways worse than the one in Medieval! Ideally I wanted to see the AI actually formulate and adopt a simple, overall strategy for each battle, much like the tactical AI in the Combat Mission games where a 'commander' level AI would set the tone and basic strategy for the coming battle and direct the units in a sensible manner until contact was made and they would fight their own battles as their individual AI routines saw fit. As Parmenio said, more often than not everything looks great for the AI until the battle lines get close and then it seemingly loses its mind and its shoots itself in the foot.

I've said it countless times before but having a reasonably effective AI opponent is the key to this game's long term success. Once the 'gee whiz' factor of the graphics and strategic gameplay/features wears off we're left with a game that doesn't hold up well, even after extensive modding. To me the Total War games were ALWAYS about the battles with the strategic gameplay coming in second.

[DisruptorX]

Suicidal Generals wasn't a problem in the version of Viking Invasion I bought, In fact, there is no AI control for your units in MTW. Did I miss an older version of the game or something? Player has complete control over his units in MTW, in fact, get this: when you order a unit to do something, it does it! (the concept is amazing, CA really needs to get in touch with the guys who made MTW and get them to help them to apply it to Rome)

[Red Harvest]

DisruptorX,

You misunderstand. The AI still controls the enemy in STW, MI, MTW, & VI...and RTW of course. In early versions of each generals were very prone to charging to the front to die. This was toned down somewhat with VI. It is back with a vengeance in RTW. The AI needs some sort of rules to tell it not to engage the general unless various criteria are met such as: 1. There can be no effective retreat. 2. The force of his unit can decide the battle and the risk is acceptable. 3. All/most other units have already engaged, and his unit is needed. 4. He should choose to engage where he won't be plowing into pikes head long, etc.

However, let's all not forget that MTW/VI was bad about keeping multiple heirs in a stack, so that whether or not the King died, he might lose all, most or several of his heirs...even in a winning battle. That was a strategic AI blunder. It still exists in RTW, but to a lesser degree. The only time it made sense in MTW/VI was when the heirs loyalty was so low that they had to be kept in the same stack to prevent civil war, or if the faction had lots of excess cruddy heirs that needed to be "cleared out."

[DisruptorX]

Ah, well I found that the enemy general in MTW was often able to retreat successfully after losing the assault (more than 50% of the time the enemy general lived). In Rome, I kill their general EVERY time.

[HopAlongBunny]

I seen the AI form reasonable battle lines. I've seen it deploy flankers fairly well. I've even see it use skirmishers correctly on the odd occassion.

But ever time it goes for the big melee crunch it just loses it. I've seen solid lines of phalanx suddenly charge apart in different directions losing all cohesion. It'll recklessly charge any cavalry (not just generals) after skirmishers regardless of how many heavy units are packed together right behind them.

Actually, you would see exactly the same thing happen with your armies if you targeted for each unit in your line...and didn't change targeting as units shift. I even did it myself a cpl times when I started playing

The difference is we get smart and update/modify our attack as conditions change. The "suicide" generals would probably just say: It seemed like a good idea at the time

[troymclure]

it's not a bug. It's just realistic.... every 3rd general wants to be the next alexander the great so attempts to emulate his "tactics" whenever possible.
It's going to make the alexander expansion pack easier as well... no need to code new AI for generals.

[Rosacrux redux]

A tad bit concerning historical accuracy: The Roman generals never fought in the first line. So much is true also for all (bar none) "Easteners". And for the Karthagenians too - they wouldn't commit to battle unless everything was lost - Hannibal himself, for instance, had escaped more than one crushing defeat (to die off poisoning later...) by simply not being among his troops.

On the contrary, of the "civilized" factions represented in the game (since the "barbarians" are a completely different story), the Greek leaders always led the battle from the first line, and in the most dangerous position possible (if in a hoplite phalanx, that would be the right wing).

That is the reason that in Greek warfare the generals usually are among the victims of any battle - they always led from the first line, that's the Greek way.

The Greeks are like that (we still are, actually): we don't like going to battle without being led directly by our supreme commander. We want him to share our fate, and put his arse on the line, just as we put ours.

It's... a Greek thing, I think.

[Parmenio]

Actually, you would see exactly the same thing happen with your armies if you targeted for each unit in your line...and didn't change targeting as units shift. I even did it myself a cpl times when I started playing

The difference is we get smart and update/modify our attack as conditions change. The "suicide" generals would probably just say: It seemed like a good idea at the time

That's true more or less. However I've been charging (or advancing to connact) my whole battle line on mass by ordering it to move to a point behind the enemy line since STW. (With Legions I set to fire at will and charge the line to just before the enemy line to trigger pilums first.)

[Red Harvest]

Yes, the breakdown of the lines will happen if you or the AI targets individual units. What happens is that if the targeted unit turns or moves, the "lead computing" is done to plot an intercept course. This makes a mess of the lines since both the human and the AI end up with a bunch of odd vectors.

As humans we are learning to overcome this problem. With phalanx units I order my entire battle line to "march through" the enemy battle line. I do the same with some other infantry. When contact is made I take control of individual units to redress the line or halt/turn/flank/etc. as needed. CA needs to program the AI to do this as well. Presently the AI battle line is prone to disintegration. However, it does a somewhat better job of meeting mine if I do the "march through" because the lead computing is nice and symmetrical.

There are some problems with group marching that can cause faster units such as cav, etc. to advance in front of the heavy infantry, even if they were grouped behind (another area CA needs to address.) I think this is causing a portion of the suicide daimyo behaviour. Still the AI should not be charging the general forward in the first place.