about smarter general AI

[DrAzTiK]

Hi David,

 

As with SCS, enemies above a certain minimum intelligence will usually prioritise opponents who are not stunned, held or otherwise helpless. From their point of view, they're fighting to win, and if they kill all non-helpless opponents they can finish off the helpless ones at their leisure.

 

From my experience, this feature advantage the players in many instances (maybe 80%). I have saved many reload thanks to it.

The more you play at middle-hight level, the more you have some stun/held effects that last only 1 or two rounds ( from items) and in these cases, it's really an error for enemies to not try to hit helpless opponents .

 

Again at middle-hight levels, priests/mages have so much memorized spells allowing to cure most of stunned/held effect that we usually don't have many characters helpless for a long time.

 

But surely the things are different considering you are playing Bg1; BG2 or TOB. I didn't make some test but surely this feature is nice and better for BG1.

 

 

 

 

Also, I think enemies should avoid to hit Mordenkainen's Sword.

[DavidW]

This is a good point. Certainly that feature originates from the BG1 version of SCS and moved into the BG2 version largely unchanged. I'll give it some thought.

 

Having said that, here are a couple of counter-arguments:

 

(i) Although on average it might be more sensible at high level to attack helpless opponents, it's almost never going to be really stupid to attack non-helpless opponents. The reverse might well not be true. My top priority with SCS is not making the AI fight as well as possible; it's avoiding situations where the computer does something so stupid that it breaks immersion.

(ii) I worry that attacking helpless opponents significantly increases the probability of having to resurrect someone without making that much difference to the difficulty. Resurrection is a bit of a pain (collect equipment, sort out spells, etc) and so I'd rather not just spitefully kill off PCs if it's not going to make a real difference.

[amanasleep]

This is a good point. Certainly that feature originates from the BG1 version of SCS and moved into the BG2 version largely unchanged. I'll give it some thought.

 

Having said that, here are a couple of counter-arguments:

 

(i) Although on average it might be more sensible at high level to attack helpless opponents, it's almost never going to be really stupid to attack non-helpless opponents. The reverse might well not be true. My top priority with SCS is not making the AI fight as well as possible; it's avoiding situations where the computer does something so stupid that it breaks immersion.

(ii) I worry that attacking helpless opponents significantly increases the probability of having to resurrect someone without making that much difference to the difficulty. Resurrection is a bit of a pain (collect equipment, sort out spells, etc) and so I'd rather not just spitefully kill off PCs if it's not going to make a real difference.

 

I would definitely say that killing helpless PC's makes high level battles more difficult, since the player loses their abilities for the rest of that battle (or gimps them due to using resurrection and not having armor/spells, etc). I think the OP is claiming, quite correctly, that the current behavior allows disabled characters to be revived at full strength and therefore make a material contribution to the current battle.

 

I find it hard to think of a situation where this behavior would be immersion breaking. Since almost all of the enemies are evil in BG2, focusing on the weakest, even if doing so were a tactical error, is a classic "evil" move.

[DavidW]

I find it hard to think of a situation where this behavior would be immersion breaking. Since almost all of the enemies are evil in BG2, focusing on the weakest, even if doing so were a tactical error, is a classic "evil" move.

 

I found it happening fairly frequently in vanilla BG1. In that setting, disabling effects tend to be difficult or impossible to remove, so someone who's, say, asleep or held is mostly out of the fight. The AI would find itself hacking at helpless foes while weak-but-fighting people were attacking it.

 

I'm somewhat inclined to weaken this a bit in the next release. But I continue to worry about my (ii) above. I also find it a little unrealistic for a monster to hack away at helpless person A while being hacked upon in turn by non-helpless person B, rather than have to defend themselves against their attacks. (I know the player can do this.)

 

At the moment, SCS AI will ignore even a close paralysed opponent in favour of a distant but active opponent, so that the priority order goes

 

1) close, active opponents

2) distant, active opponents

3) close, helpless opponents

4) distant, helpless opponents

 

I might try swapping 2 and 3 (while keeping 1 at top).

[Dakk]

Your (ii) point is the one I find the most compelling, and I agree. Not killing of PC's just "because you can" if it's not changing the actual outcome or difficulty of the battle is laudable. I like your proposed change to targeting priority. If I understand correctly this would mean that a "tank" PC held by a ghast would get wailed upon until reinforcements come up and start beating the ghast - because then the ghast would switch to the active assailants? This would be very desirable behaviour.

 

And like you say, it is actually quite common (in BG1) for a PC to be held/stunned and slowly be chipped away at while the rest of the party is incompetently trying to beat down the surrounded monster. It feels like a cheap death/resurrect.

[DavidW]

Your (ii) point is the one I find the most compelling, and I agree. Not killing of PC's just "because you can" if it's not changing the actual outcome or difficulty of the battle is laudable. I like your proposed change to targeting priority. If I understand correctly this would mean that a "tank" PC held by a ghast would get wailed upon until reinforcements come up and start beating the ghast - because then the ghast would switch to the active assailants?

 

That would be the idea, yes.

[Dakk]

Your (ii) point is the one I find the most compelling, and I agree. Not killing of PC's just "because you can" if it's not changing the actual outcome or difficulty of the battle is laudable. I like your proposed change to targeting priority. If I understand correctly this would mean that a "tank" PC held by a ghast would get wailed upon until reinforcements come up and start beating the ghast - because then the ghast would switch to the active assailants?

 

That would be the idea, yes.

Great!