Should encounters have aggro mechanics?

[ledroc]

I like the idea of enemies that vary in perception and intelligence and choose targets based on a variety of factors, including both proximity and who is hurting them most.

 

THIS is the intent of the poll, which is essentially "aggro"

 

How does the AI decide what character to target and make that dynamic during the fight? That is the real question

 

What math determines that?

Is it random?'

Is it based on damage?

Is it based on perceived threat?

Is a caster in the middle of casting a spell and should the enemy interrupt that spell?

 

All of those actions generate "threat" or whatever you want to call it and the AI should react to those factors

[Jarmo]

The first part yeah, the second NO!

 

Yeah, I'd like them to switch targets sometimes, to whoever damages them.

Hit them with an axe, yeah, that should get their attention!

 

But no, I don't want my fighter to start "generating aggro" so he's attacked.

 

Would you like some damage sponge enemy suddenly starting to generate aggro,

so you can't choose who you attack but have to start spanking the sponge?

[metiman]

How would you determine it, ledroc? Would you go aggro on the guy with the highest hitpoints and the best armor while seeing 3 powerful spellcasters behind him with almost no hitpoints and no armor? That fighter is going to hit you no matter what you do, but you are not going to be able to damage him all that much compared to what you can do to the casters: kill some outright and interrupt at least some of their devastating spells. I've never understood strategies that don't go for the casters first. And, yes, that is what Sword Coast Strategems does. If paying attention to "aggro" mechanics were smarter those BG2 mods would have options for that. They don't because aggro mechanics are for foes with an IQ of like 3. Well except that it wouldn't work on them either.

[Shadenuat]

I would like to see attitudes and behaviors. Something with more variety. Maybe Barbarians hate mages and all beeline for mages. Maybe another type of enemy likes to prioritize party members in heavy armor.

 

I like that. I think stupid monsters should actually hit the thing closest to them, and die in traps and spells. On the contrary, intelligent enemies well versed in art of adventure should pick your party apart starting from support characters, leaving slow fighters in a Web spell. Something like that.

[Vampero]

Wrong kind of game for that. This game is not a MMO.

[ReyVagabond]

Well yes and no, in a party enviorement the enemy needs to have enogh inteligence to kill whats its most anoing for them.(Its all about the enemy AI).

 

What i dont whant is a gimic Tank and spanc mechanic, and or a 40+ Taunts to keep the warrior the senter of the enemies atention.

 

I want some real Enemy AI that can take desition and despite a taunt can distract an anemy for 2 secs does not change his objective. if he cant hit a rogue then he goes for another target, i want the AI to be good.

[el pinko grande]

This poll really needs to be reworded. What people really object to, I believe, is MMO-style tanking based on taunts. Which yeah, is lame. But there definitely needs to be a threat/aggro system where the enemy AI reacts dynamically to whichever target they perceive as being the biggest threat. And it should take account of more than just who did the most damage to the monster most recently. If your healer is throwing out massive heals, he probably should be a target, yeah. If the mage keeps casting Hold Monster on his buddies, then the mage should become a preferred target.

[ledroc]

How would you determine it, ledroc? Would you go aggro on the guy with the highest hitpoints and the best armor while seeing 3 powerful spellcasters behind him with almost no hitpoints and no armor?

 

No, but I wouldnt allow some archer sitting in the back constantly shoot arrows at me either. There just needs to be some better intelligence built in than most games currently have.

 

Good graphics are easy to program, good AI is not

[Lady Evenstar]

I've never understood strategies that don't go for the casters first.

 

Doesn't that depend on the intelligence and abilities of you and your enemy? Better to do your part by chewing on what you can reach than dying en route to a caster, no? Unless it's specifically your job to focus the caster's attention, I'd say you were better off eating a sacrificial thief than letting your enemy pick you off as you run toward a ranged target.

[PsychoBlonde]

Just wondering what people thought about having some type of aggro (aggression) table in the game.

 

There's going to be one regardless--enemies have to have SOME kind of formula for deciding which character to attack. What there may not be is abilities that directly affect aggro--so the only way to get it or keep it may be to be the first character in sight or to do the most damage. I'm fine with that--party-based games are rarely designed so that ONE character can safely eat ALL the aggro.

 

I also don't mind if they have enemies who ALWAYS default to aggro on casters--this could be a feature for higher difficulties, even, where enemies will ALWAYS use their abilities on the character who is (theoretically) weakest against them. You could also have mechanics where some classes (fighters, perhaps) can choose to soak damage/spells directed at a given other character.

[Gurkog]

No aggro means melee characters are a waste of space since magic users will always be useful while they have to chase stuff around to MAYBE get its attention. I guess you would just have 1 wizard kiting the enemies while others are blasting away on them. If they make the enemy AI as well designed as The command options you can give in DA:O (hopefully more complex than that) they can roughly simulate logical tactical decisions. I rarely had to give manual commands to my ranged and support units because they did what I wanted automatically.

 

There seems to be a lot of beligerant knee-jerking in this thread that would make the rpgcodex proud.

 

EDIT: My D&D tactic for BG and IWD for dealing with finicky threat is to summon an impenetrable wall of creatures to block the enemy from attacking my ranged units. Incredibly cheap and incredibly effective.

Edited October 4, 2012 by Gurkog

[PsychoBlonde]

Good graphics are easy to program, good AI is not

 

Actually, this is only sorta true. What's difficult to program is an AI that is smart enough to challenge the player but ALSO dumb enough to let the player win at least some of the time. Since players come in so many, many different levels of skill, too, they usually have to err on the side of stupidity. This is the GM's problem in pen and paper, too--if the GM wants to win, they can. They can have every monster go straight for the healer and wizard like gangbusters. They can have all the archers shoot ONE person (the fact that they'd do this in BG was one of the things that made groups of archers so horribly nasty). The GM can have the Evil Wizard scry on the party (the only spells that prevent scrying altogether are location-based, useless while you're actually adventuring) until they are low on resources and need to rest, then use Greater Teleport to dump a ton of monsters on them.

 

After all, players do this stuff. I've had players summon 100 lantern archons or arrowhawks to kill a boss--each may only do a little bit of damage, but their primary attack is a ranged touch attack of a type that basically nothing resists. If you have them spread out, even the biggest AOE spells won't get rid of a substantial number. This is also part of the reason why at higher levels (anything past, say, a single big boss is a joke fight in D&D--there are too many ways to cheese them out.

[Ninjamestari]

The threat mechanism might actually work if done properly. For one, the threat levels should represent the combat situation, so you couldn't build up threat in advance in order to make an enemy completely ignore a much deadlier foe that enters the battle. Maybe the threat levels could be tracked for the last 5 seconds or so to make your foes react into changing situations better. Distance should also be a factor; especially melee oriented fighters should prefer to attack enemies that are close to them. An archer might try to avoid the attention of your melee fighters by focusing on your supporting characters and spellcasters. Critical hits with ranged attacks might cause some enemies to hunt down the archer (an ogre might be rather mad after getting an arrow to one of his eyes, trying to direct his anger to the party member responsible) or things like that.

 

All in all the goal is to make enemy NPCs behave more intelligently (not all of them, a stupid ogre should fight like a stupid ogre, not like a master tactician), and more realistically.

[Adauli]

When I think at Aggro mechanics I allways have to think towards "trains" in old Everquest where people try to flee but instead of geting away they get more and more oponents following them and in the end they and a lot of other players get killed.

Or the Taunt ect. actions so that a tank can keep the agro in MMOs. Both things I hope to not see in singleplayer games.

Thats why a clear No from me to this.

But I'm all for it if there is some kind of more inteligent AI behind the inteligent opponents but there should be different AI reactions on the same situation depending on what race the opponent is and its stats.

[Vaeliorin]

I can't imagine anyone would object to the AI being smarter than it was in the IE games, where everything basically just attacked the first thing that hit them until it was dead. That said, there also needs to be some way to protect squishier members of the party. That doesn't mean taunts, and it doesn't mean convenient choke points everywhere (which is as bad as taunts when it comes to things not making sense.) What exactly it means is up to debate (stances that create a zone of control around a character that slows enemies (as they have to dodge/avoid attacks and thus can't move at full speed) or attacks them when they try to move through possibly, or trips, charges, etc.), but if enemies are going to be smart enough to charge the squishies, we need a way to keep them from being able to get there.

Edited October 4, 2012 by Vaeliorin

[Hypevosa]

I know that in my D&D campaign the wizard who pumped out 200 damage to a dragon got noticed alot faster than the warrior attempting to whack its nose off - I think the game should reflect that casters are often the most dangerous thing on the field and pumping out alot of damage. Unless there's someone inbetween to intercept the guy whose face you nearly blew off, I want to see things go after the highest threats first.

 

That said, something so irrational or stupid it can't understand the concept can keep attempting to hurt the tank.

[sesobebo]

That's just bad AI. It has nothing to do with aggro. Play BG2 with the SCS mod and you'll find that every enemy makes an instant beeline for your spellcasters just like you would.

 

Yes, that is partly my point, you needed a mod to make those games better when that intelligence should have been built in

 

Its also not just heading for the casters its being able to pull that attention away from the casters. Otherwise there is really no point to beefing up armor on a melee character

You bring them along to soak up damage

 

if that was Your point, You've presented it pretty poorly, because there's quite a few solutions to encounter AI - the MMOish aggro meter being one of the more lazy and unimaginative ones.

 

also, You can't compare AI written for a 1998 game that officially supports 166 MHz pentuim generation, to an AI that could be run on today's machines (or machines at the time SCS was made); same goes for pathfinding - which i suspect has as much to do with Your ogre vs archer encounter as the incompetent AI.

 

besides, isn't Adam Brennecke a schooled AI programmer? let him do some super special work!

[Caerdon]

"Aggro" should just be one part of all those under-the-hood calculations the enemy AI does. It shouldn't be something the player can directly influence with multiple skills like "taunt", "detach", "intimidate" etc.

[TrashMan]

I haven't read the thread, just stopping by to say

NO

I would add something more but I don't feel like getting banned before seeing the backer forum badge at least once.

 

This. KILL IT WITH FIRE.

 

With pleasure!

[Dawn_]

After having played rogue in a lot of my BG campaign and always going for backstabbing the caster (priest/mage) first..i can't answer yes to such a question..

[septembervirgin]

I don't think aggro is the right way to approach combat AI. If aggro is used, it should be modified by maybe two or three basic AI tactic sets that are each diversified and built for different difficulty levels or intelligence of the characters.

 

For example, a group of fifty bandits might have scouting in day and night cycles. The scouts they'd use would be thief class characters. The scout would not attack but would probably remain stealthed and report intruders. With aggro, scouts won't exist properly unless a part of a story. Also during battle, scouts would be taking turns attacking then hiding, never keeping an easily noticed formation. Meanwhile, the fighter class bandits would be doing more direct battle, with archers behind spearmen, and shorter arm fighters trying to hit from the flanks. Of course, archers would probably focus "on aggro" unless they thought they had a swift kill in a mid-danger target. If you can't paralyze the heavy damage dealer, restrict their ability to do damage then turn your attention to the priest. If you can't kill the priest quickly, all focus on the easiest target to kill sequential to the danger they present.

 

A wizard's tower might indeed have living guards or animate guards -- but we've all seen the failure of guards that don't operate in tandem with eachother. Guards trigger alarms, run to their emergency stations, then defend en masse, with larger groups fighting the foe and preventing retreat, slowly closing in. A wizard's tower is easier to program the AI for, as angles of approach are easier to define. The wizards could be said to be a smarter foe and would second guess some character weaknesses and vulnerabilities ahead of time, so they would "detect" the best targets ahead of time.

 

While this might sound a bit lethal for a combat AI, planning and saved games can help one get ahead in any struggle. Far more lethal AIs can be formulated.

[Gorth]

Thread pruned a bit. Don't like the idea? Fine. Don't like the poster? Keep it to yourself. Saves other people from getting 'Aggro'

[jivex5k]

Aggro can be done well, and it can be done poorly.

I'd rather see a tactically smart AI where it's appropriate, and stupider ai where it's approriate.

IE: Giant Spiders should just be hungry. Attack first target.

Groups of Kobolds with a leader should use flanking, and target priority.

How to choose that priority is a question beyond my capabilities.

[nikolokolus]

I guess for most of you that say "no" you prefer exploiting ranged mechanics and having endless kiting encounters?

 

No, I prefer enemy AI that looks for the weak squidgy bits in my party and then tries to kick those weak squidgy bits as hard as they can. If that means bee-lining for my mage, or a giant hurling a boulder at my musket wielding ranger then so be it.

[Weizor]

No, I prefer enemy AI that looks for the weak squidgy bits in my party and then tries to kick those weak squidgy bits as hard as they can. If that means bee-lining for my mage, or a giant hurling a boulder at my musket wielding ranger then so be it.

But there has to be some way for the player to force enemies to attack your defensively specialised characters, otherwise they become quite useless. Some sort of ZoC-system would be preferable but those can be tricky to implement in games with real time combat.