Trending Games | World of Warcraft | Overwatch | Anthem | Entropia

    Facebook Twitter YouTube YouTube.Gaming Discord
Quick Game Jump
Members:3,836,149 Users Online:0

Show Blog

Link to this blogs RSS feed

In My Humble Opinion

My ramblings on where the MMO's of tomorrow are going to have to go to keep us all interested...

Author: Melf_Himself

Revisiting the Holy Trinity

Posted by Melf_Himself Tuesday April 29 2008 at 4:31AM
Login or Register to rate this blog post!

I've read a couple of blogs today about the Holy Trinity. First was heerobya's:

Heero's position was that the trinity isn't THAT bad, it leads to camaraderie / dependence on your fellow player.

Next vajuras responded with:

His position was that it's unrealistic for an enemy to attack the most highly armored target on the field. The enemy should always target the healers.

I've touched on this in my previous blog, but it was loooooong, and I'd like to make a couple more points along with a possible foundation for a combat system, so:

1) PvP and PvE players are always segregated in MMO's. This is because completely different tactics are used. This is because monster AI is nowhere near as smart as human AI.

Monster AI often attacks the nearest player, not the most vulnerable, it doesn't learn from its mistakes (ie continues to attack the tank even though it's taking no damage), and it often doesn't run from AoE effects. Designers also gimp monster groups, usually not allowing them to have a good mix of monster classes.

Because monsters are so dumb, they are invariably buffed in either damage or hit points or both, in some cases to flabbergastingly astronomical levels.

2) It's not as simple to fix the AI as to just "make it go for the healer". This is for 2 reasons:

a) If the healer being targeted is smart, it will see the monsters have targeted him, and run away. This is called "kiting". The AI then needs a further step, to identify when their target is kiting, and to not chase them too far before switching targets.

b) Usually the best target to pressure are actually the NON-healer casters. Think about it: if you attack the DPS, you can either interrupt it or force it to kite. Either way, you stop it from doing its job (doing DPS). Whereas, if you attack the monk, the monk will kite.... but it's still doing it's job by kiting (it's job is to negate damage). And, that DPS guy that you COULD have targeted, is still doing his job.

So, by attacking the healer, you don't achieve anything - if the healer is smart. And we want our AI to be able to play well against smart players.

The AI needs to be able to have options open to it to disrupt enemy players, and to be smart enough to disrupt the right players at the right times.

3) In vajuras's blog, JB commented that the AI shouldn't be always programmed to do the same thing under the same circumstances, do decrease predictability. I agree with this to a certain extent. Many theories about having "fun" in games state that the fun comes from analysing patterns, and solving problems based on those patterns.... if the AI is to predictable OR too unpredictable, the game would likely not be as much fun.

However, as long as the AI does whatever a smart player would do, the AI is predictable to an extent (you just have to know what a smart player would do). This would allow people to decipher the pattern.

4) Following on from point 3, you don't want to always have to be wracking your brains every PvE fight. The AI needs to be scalable. You could make this tie easily in with realism - eg the horde of zombies you fight in the evil liche's castle use stupid tactics, but the liche at the end with his posse of uber zombies and vampires are smart, and consequently tougher to take down.

Another option is just to make a gradual transition as players move through the content in terms of the AI slowly getting better, with the game completely finishable where the enemies only use moderate AI, and optional "end-game" content with "real" AI.

5) Where do warriors fit in under such a regime you ask? vajuras mentioned that as long as server-side collision detection is implemented, the warriors can still deter enemies from attacking their squishy casters. However, another poster raised the issue that this would cause a large imbalance in the mage : warrior ratio of player characters needed, since you could have one line of warriors with a large group of mages behind.

An idea to stop that from happening:

Make warriors the best melee characters

Make the best disruptive characters in the game need to go through warriors to be able to disrupt the mages.

For example, you could have the assassin who's abilities may be centred around throwing casters off balance, or you could have a DnD monk-style "anti-magic" character who is able to cut through magical defenses, etc. You could also have ranged disruption in the form of archers/rangers.

You'd then have a dynamic situation in which the mages deal large damage but are interrupted easily by assassin/monk/ranger types, and so require warriors to protect them from these types, who could in turn be temporarily disabled by mages, etc.

You still have a rich inter-dependance between all the classes as heero said is a good feature, but it seems to make a bit more sense than ye olde Holy Trinity.

That's just one idea - I'm interested to hear any shortcomings people see with that system, and especially any other ideas people have had to end this Holy Trinity nonsense.



I haven't blogged here for a while. Last blog I mentioned a lot of things I hate about current MMO combat systems, and hoped that in my next blog I'd have magically solved all those problems. I haven't come close to solving them all, but hopefully this is a start :p



grimfall writes:

The original 'Holy Trinity' was actually Tank, Healer and Crowd Control.

There's other ways to address the 'problem', like giving your tank both ranged and melee attacks and/or making it so if you're not targetting a tank class you get exposed to critical strikes.

The main  problem in PVE isn't damage or ability based discrepancies it is distance based discrepancies.  If you have  a game with no ranged attack, then proximity could be used to guard the weaker classes.

Tue Apr 29 2008 5:33AM Report
vajuras writes:

Point #1: In PVP, if you dont catch the Healer you pretty much lose. Healer/Rezzer is always your biggest threat and must be contained. If we have a clump of 5-6 mobs then at least one of those mobs should dash for the healer. In RPG AI, mobs are already inherently coded to stop pursuing the target after they get out of range. So if the target wants to run away from the entire encounter let him. But you can never ever ever ignore the healer. Healers must always be contained and eliminated

Kiting is a valid strategy so if players want to employ its a valid strategy. Perhaps if we cant kill the healer then other rules should of course come into play.

Mages can be equipped with FireWall spells to help setup a blockade.

Warriors should be Warriors- wrecking stuff. They definitely cannot be a "Taunt" sort of role logically unless they somehow become promoted as biggest threat. But there lies another problem with that- in PVP players move around and kite Warriors. Thus they normally are least desired in PVP without the aid of snares. In Guild Wars Warriors have snares and fastest players on foot. This made Warriors viable plus they can equip Mage (long range) abilities

Point #2: The only way to really end the Trinity is to toss away "Sports gaming" completely. Trinity occurs due to optimization of team distributions. In sandboxy games there are no teaming restrictions- no party limits. Thus they avoid "Trinity" issues per se. Unfortunately, most games still have an inherent sort of Trinity thing happening behind the scenes but as long as they rigiourously pursue the healers like we see in PVP then we'll have more realistic AI

Tue Apr 29 2008 10:14AM Report
grimfall writes:

Sandbox games has nothing to do with character classes. 

Tue Apr 29 2008 11:33AM Report
vajuras writes:

I dont understand Sandbox Games inherently always generate Dynamic Classes. In EVE Online your ship is your Class and provides a level of Tactical Transparency

In Elder Scrolls you pick a generic Class template

You really need to elaborate because most any sandbox you name (aside from Second Life) does indeed dynamically generate Classes. Tactical transparency is usually almost always there in some visual format

Tue Apr 29 2008 11:37AM Report
BadSpock writes:

Even Ultima Online, the holy grail of sandbox MMOs, had defactro classes that formed just due to popular combinations of skills.

Tamer / bards. Mages, Axe + Lumberjacking, etc.

All these ideas you have all boil down to the same thing-

Make a more complicated way of doing the exact same thing they are currently doing.

If you have a single mob, that mob is going to only be able to physically be near a certain number of characters at once in order to do melee damage. If they do ranged attacks, they can only target one specific target at a time, unless it's an AoE attack or conial etc. 

So how do they choose targets? They go for the biggest threat or the most immediate threat. The most immediate threat will always be your melee because they are going to be right in the mobs face. Its like getting through the defensive line in order to reach the quarter back.

The blitzers still have one thing on their mind, get to the QB, but in order to do that they have to get through defensive line first. So you shift focus to the most immediate threat.

The defensive line have abilities and skills they have worked on to keep the blitz from reaching the QB. The QB can always run for it (kiting) if they blitz gets through right?

But this is a game, not real life. Having collision detection is a good step for the linebackers (warriors) to keep the blitz from getting to their QB (cc/healers/dps) I'm not going to argue there. But if you don't have collision detection, you need something else. So you use a skill like taunt. Which pretty much means "hey, I'm right in your face kicking you in the groin, focus on me not the old-man-wizard back there."

Throw in crazy MMO speak things like aggro drops, shifting threat tables, make aggro specific for the mob (like it is in WoW) and all of a sudden you have a very simple concept where you add tons of simple rules that make it not-so-simple anymore.

If you actually sit down and list all the things that can/could/will/do happen in a "typical" pull in a Heroic instance in WoW or in a boss fight.. it's stagering how complex it is. Sure, it's going to be mostly the same each time, but I stil believe that is a good thing.

PvE is designed for the players to win. Plain and simple. That level or predictability is a GOOD thing. Removing it would be a mistake. Everyone would be running all over the place dying a lot and yelling/screaming at each other, no one would have a set job or role so everyone would end up trying to do many things at once...

Pretty much, it'd be a cluster-f*ck.

The holy trinity adds a level of cooperation and coordination and gives the potential for leadership to emerge.

What happens if you take it away? You'll end up doing the exact same thing-

"Ok we know that they'll go for the healers and clothies first, so let's set up flame walls here and here and then put the tanks here so we can keep the baddies from getting through, if something breaks through this person try and kite it over here etc. etc."

Which is pretty much the EXACT same conversation you have before a complex pull in a holy-trinity based game like WoW.

Tue Apr 29 2008 3:45PM Report
Hexxeity writes:

You could always just change how your classes are set up.  Not do away with them, just make them different.  As in not based on melee-only, DPS-only, and  support-only.  If roles can be assigned in a meaningful way, but a way that does not fall along the Trinity lines, then Poof!  No more trinity.

How to set up those classes is, of course, a very lengthy discussion for another time.  And there is the real possibility that a new trinity could arise.

Basically, heerobya hit it on the head when he said PvE is designed for the players to win.  As long as this is true, you will always get complainers saying the game is too easy.  You can make a game where it's not true, but that's very risky, and you'd have to be sure the players are aware that they are not always going to win if they go up agains the smarter monsters.

Tue Apr 29 2008 5:25PM Report
vajuras writes:


The holy trinity adds a level of cooperation and coordination and gives the potential for leadership to emerge.

What happens if you take it away? You'll end up doing the exact same thing-

"Ok we know that they'll go for the healers and clothies first, so let's set up flame walls here and here and then put the tanks here so we can keep the baddies from getting through, if something breaks through this person try and kite it over here etc. etc."

Which is pretty much the EXACT same conversation you have before a complex pull in a holy-trinity based game like WoW."


holy trinity is good for simple, sports gaming. But true level of teamwork we see in EVE Online is crafters making gear for alliance, recons out there reporting Intel, miners out there mining Ice, Traders/Haulers bring us resources.

Where is trinity? Its gone. Imagine other scenarios besides simple Dragon killing. And trust me- we work together much more close knit then you ever will in a "scheduled" raid. We deal with dynamic challenges. You raiders perform dance routines



Tue Apr 29 2008 8:46PM Report
grimfall writes:

Your level of ignorance concerning PVE raiding is quite astounding.


Crafters making gear for alliance = Yup

Recons out there reporting intel = Yup

Miners out there mining ice = Don't know what that means... is this for drinking water?  But if it is farming resources, that's a Yup

Traders/Haulers bringing us resources = Yup

What any of that has to do with the actual combat encounter (except the intel) is beyond me, though.

Wed Apr 30 2008 6:57AM Report
BadSpock writes:

The teamwork you speak of in EVE online is great, yes, but do you believe that similar doesn't happen in a raid centric game like World of Warcraft?

You have the leaders who organize and recruit in order to get the people together of the right skill levels, cooperation, and class balance. This takes a comination of organizational skills, people skills for recruiting, diplomacy skills to keep things civil etc. 

You have the crafters who make more then just weapons and armor for themselves, but create potions and enchantments and flasks and elixers etc. to aid the group.

You have the "scouts" who have been there before who know the situation, the strategy, or those who research and plan the encounter down to the finest detail. It may not be in-game scouting as much as EVE, but it counts.

Then during the actual encounter, well if you've played through any raid in WoW you'd understand that level of cooperation, coordination, timing, communication, and yes, even player SKILL involved.

I'm not trying to mock you Vajuras, I just think you're a little too defensive about EVE and not open minded enough about other gamaes.

I admit that I'm no EVE expert, by any means, but I try to keep an open mind.

All the "holy trinity" really is, is a way to say "specialization." You say that in EVE you have crafters and scouts and miners and traders etc. It's the same thing, its specialization. 

During combat don't you have different specializations? Tanks, nukers, warp jammers, etc. etc.

Wed Apr 30 2008 11:25AM Report
grimfall writes:

During combat don't you have different specializations? Tanks, nukers, warp jammers, etc. etc.

Yes, but it's dynamic.  You'll never understand what dynamic means unless you play Eve 0.0.  Sorry.

Wed Apr 30 2008 8:53PM Report
Melf_Himself writes:

I think the point has gotten a little side-tracked by the whole "skill-based" vs "class-based" thing.

If the AI is simple in a skill-based game, the best strategy to fight it will be the same - which will be to min/max your skills to form "classes" to do particular jobs.

So whether you go skill-based or not, the first thing that would need to be done is to improve the AI. Then you can have whatever type of game you want (class or skill based). I think that's the real issue folks.

@ grimfall: I see what you mean about proximity vs distance. It would be useless to have melee defending your squishies if all the DPS is coming from ranged anyway.

My attempt to overcome that was to basically make it so that, although ranged DPS would dominate, it would require disruption of the enemy squishies in order to be effective. And that disruption would be melee, hence giving the warriors a purpose, and adding some dynamic strategy in terms of positioning.

Whether the whole thing would work or not is arguable, it was just one idea. I hope others think about it and come up with a much better solution, I know I shall continue to mull it over ^^

@ vajuras: Warriors "wrecking stuff" - ie being the best DPS - is a good solution, it is after all the way that Guild Wars does it. Although it turned out to be a balanced game, I just always kind of had a problem with the fact that it felt so weak to be a mage, it seems to run counter to standard fantasy lore.

The only time in that game you felt "uber" as a mage was when you played in PvE vs stupid AI that lets you farm them in AoE... again, bringing us back to the whole AI problem.

@ heerobya: I know what you're saying about PvE encounters being designed to let the players win. You definitely need some less mentally intensive encounters to please the majority of newer players.

But I think at the same time, you need the option to go "hardcore" for your established vets. Having optional areas not crucial to the main storyline where you go up against smart enemies would help keep the game alive. In order to make new players not feel like they have to go there, make the rewards for it more in the form of bragging rights then power. So drops would stay the same but they could work on a special title, or get points towards a unique looking suit of armor, etc.

I know WoW high level raids can be difficult, but it's mainly trying to get people to do what you tell them to do and the general extreme challenge of getting 40 people all online wanting to do the same thing. The combat itself, once you've memorised the routine by rote, is not really that tough.

The reason I really want to see some kind of "advanced AI mode" implemented is because currently, when people tire of end game PvE content and try PvP, they inevitably suck. Which makes NO sense, they dominate at a tricky aspect of the game, and then completely fail at another. The reason for that failure should be laid squarely at the hands of the AI.

I know some people who hate hate hate PvP, and for 90% of those I'm sure that if they owned at it, they would like it (some just don't like the idea of attacking another person, which is fine, but I think that most are just scared of getting owned/laughed at). If PvE prepared them, *gradually*, for the jump to PvP, we could all have the PvEvP RvR epic scale wars that I think we've all been dreaming about here at for some time!

Thu May 01 2008 8:12PM Report
HumbleHobo writes:

PvE needs to be more like PvP.   I fully agree that the best way to do this is to improve the AI.


More diverse mechanics can also help make strategies viable in both categories.  Like ambushing and scouting.

Fri Aug 01 2008 8:02PM Report writes:
Login or Register to post a comment