I was reading this great article about building your player constructs through dynamic components(which is technically neat but way overkill for what I'm working on).

It got me to thinking about the same concept in "class" design.

The two well known approaches to this are to either have a class-based or a skill-based system.

There are certainly other concepts which attempt to hybridize these two systems, and this one is likely of that sort.

Briefly, the common wisdom is that a class based system is easier to balance and can ultimately provide more diversity, as there tends to be a "best combination" of skills that everyone ultimately migrates to with a skill based system.

The downsides of classes are that they can feel restrictive, and tend to homogenize the player base(although, if your skill based system ultimately merges into one set of skills it will still be more diverse)

The final selling point of classes tends to be that they increase replayability, assuming the experience is varied enough from class to class, and the content is compelling enough for you to want to run through it again with a different set of skills.

The big issue with skills tends to be that it is difficult to balance, and can lead to "one build to rule them all" issues. It also tends to break the holy trinity of tank/healer/crowd control that most cooperative rpgs are based on.

Classes major shortcoming tends to be extreme specialization that makes for "dud" builds when solo. The super healer can't kill anything, the super-tank can't either, a dps class can neither tank nor heal at a raid level but does great solo.

Consider a skill based system that allows you to choose a series of "components" or "kits"

To develop this idea I'm going to start with a common archetype, a "warrior" at level one.

At level two and beyond, this guy can run straight up the warrior path until it branches into the standard, tank and dps routes.

What if your choices were diverse enough where by mixing and matching components you were effectively a different class, and each of those classes excelled at tanking a particular scenario of mob?

The idea I suppose ultimately boils down to a skill system, so there's not enough meat to the idea really.

I guess the better question may be, how can you reward diversity of skillsets and not have the "one build to rule them all" issue, while at the same time not thoroughly breaking roles?

The one thing classes do very well is to provide an easily understandable snapshot of what a charcter may do. A warrior can tank a priest can heal. Talent tress in WOW for instance muddies that up a bit, but imagine having to have a 3 minute dissertation when you ask someone if they can tank ad they have to explain their build to you.

Quote:Original post by Dreddnafious Maelstrom
The big issue with skills tends to be that it is difficult to balance, and can lead to "one build to rule them all" issues.

The "one build to rule them all"-issue is something you can't get rid of. Whenever you give the option to customize your character (by skill, skill trees, components, subclasses), someone will find the optimal configuration for a certain task. The best farming build, the best pve-orc build, the best pvp-anti-mage build, whatever. The abusing of the best-X build depends only on the player. There're players who will always choose whatever they feels right, even if it is not optimal (I'm one of these players), there're players who want the optimal configuration (a friend of me) and there're players who just copy the most hyped configuration from community forums. The latter one is the reason we often see the "one build to rule them all" -myth :-)

Quote:
Consider a skill based system that allows you to choose a series of "components" or "kits"
..
What if your choices were diverse enough where by mixing and matching components you were effectively a different class, and each of those classes excelled at tanking a particular scenario of mob?

Many games already try to archive something similar. There're the sub-class faction (Drageon Age, EQII),the skill tree faction(WoW, Diablo),the component faction (Fable?), the slot/build faction(Guild Wars).

Quote:
It also tends to break the holy trinity of tank/healer/crowd control that most cooperative rpgs are based on.

The holy trinity of tank/healer/crowd is just a EQ relict. Take a look at older,class based RPGs or Pen'n'Paper RPGs. These old games didn't support the of tank/healer/crowd(/damage dealer) to this extends. A mage was mighty and was able to fight off other creatures even in close combat (think of gandalf vs balrog), even without crowd control it was possible to fight a larger mob. Once the class decides how you have to play a character, in modern MMORPGs it decides what role you have to play.

Quote:Original post by Ashaman73
The "one build to rule them all"-issue is something you can't get rid of. Whenever you give the option to customize your character (by skill, skill trees, components, subclasses), someone will find the optimal configuration for a certain task. The best farming build, the best pve-orc build, the best pvp-anti-mage build, whatever. The abusing of the best-X build depends only on the player. There're players who will always choose whatever they feels right, even if it is not optimal (I'm one of these players), there're players who want the optimal configuration (a friend of me) and there're players who just copy the most hyped configuration from community forums. The latter one is the reason we often see the "one build to rule them all" -myth :-)

If the game supports enough diversity in what you do on a daily basis you could at least have a respectable range of "best skillsets". Going back to my tanking examples, one way to really challenge the min maxers would be to have a "best tank" build for different encounters. Like one build is a great AOE tank but another is a great anti-magic tank. Still another is the best dodge tank while the other is the best effective health tank.

The danger with that is it would be hard to engineer for starters, and even then if the game were challenging enough min maxers would probably build those specific characters for that specific encounter rather than just going with the flow.

It's a tough issue to solve.

Quote:
Many games already try to archive something similar. There're the sub-class faction (Drageon Age, EQII),the skill tree faction(WoW, Diablo),the component faction (Fable?), the slot/build faction(Guild Wars).

Sure, but EQ2 is really deferred handcuffs. You start as a warrior but then have to choose your nex class at X level. In WoW you can be a tank a fury warrior or an arms warrior but you can never sample from the rogue tree or the priest tree.

As I'm considering it, kits should at least give you a similar freedom as DDO multi-classing. Again, I'm trying to think of a system that rewards diversity as much or more as specialization, and also trying to consider the details the game would have to provide to make this possible.

Quote:
The holy trinity of tank/healer/crowd is just a EQ relict. Take a look at older,class based RPGs or Pen'n'Paper RPGs. These old games didn't support the of tank/healer/crowd(/damage dealer) to this extends. A mage was mighty and was able to fight off other creatures even in close combat (think of gandalf vs balrog), even without crowd control it was possible to fight a larger mob. Once the class decides how you have to play a character, in modern MMORPGs it decides what role you have to play.

I can't agree with this at all. Old D&D mages couldn't stand up to a basic skeleton swing for 3 full levels. There was no real aggro concept because even mediocre DM's are far more sophisticated then the best of AI's. WoW, which is the defacto authority on all fantasy RPGs(like it or not) is entirely role based. Shadowcraft, Warhammer, hell even GURPS had a tank and healer concept.

It's a very useful abstraction.


**edit, I enjoyed your journal on random quest building. I am working on a similar system for random level generation except it's for outdoor worlds. It's still a basic graph like yours except I populate the nodes with pre-determined "stages" of content and of course I have to have some of them empty so the content distribution isn't regular. Good stuff.

Quote:Original post by Dreddnafious Maelstrom
If the game supports enough diversity in what you do on a daily basis you could at least have a respectable range of "best skillsets". Going back to my tanking examples, one way to really challenge the min maxers would be to have a "best tank" build for different encounters. Like one build is a great AOE tank but another is a great anti-magic tank. Still another is the best dodge tank while the other is the best effective health tank.

The danger with that is it would be hard to engineer for starters, and even then if the game were challenging enough min maxers would probably build those specific characters for that specific encounter rather than just going with the flow.

It's a tough issue to solve.

It is my belief, that we need to design gameplay and features in multiplayer games around players and not around design dreams. The player, or even more powerful, the community, is the driving force of abusing gameplay and design features. The FPS genre has a longer and tougher history of abusing gameplay and recent FPS games start to change their gameplay to hinder players to abuse their game system. MMORPGs is just the same matter, with the issue, that MMO development is more tardy.

Take a look at modern warfare(FPS). You are able to choose different "skills" and equipment to custominze your character. This leads to differnt gameplay (the faster closecombat spec, the midrange, the sniper, the sneaker), but once a gameplay style has been choose, the best skill/gear set which supports this gameplay will be taken. The issue with the current state of MMORPG is, that you have very similar gameplay over all roles (selecting target, pressing button combo in constant pace). That leaves only the skillsets which will influence the effiency of the character.

Quote:
Sure, but EQ2 is really deferred handcuffs. You start as a warrior but then have to choose your nex class at X level. In WoW you can be a tank a fury warrior or an arms warrior but you can never sample from the rogue tree or the priest tree.

As I'm considering it, kits should at least give you a similar freedom as DDO multi-classing. Again, I'm trying to think of a system that rewards diversity as much or more as specialization, and also trying to consider the details the game would have to provide to make this possible.

You should take a look at GuildWars. It supports frequent reskilling and multi-class support (Warrior-Necro), still the issue of best-builds is present.

Quote:
I can't agree with this at all. Old D&D mages couldn't stand up to a basic skeleton swing for 3 full levels. There was no real aggro concept because even mediocre DM's are far more sophisticated then the best of AI's. WoW, which is the defacto authority on all fantasy RPGs(like it or not) is entirely role based. Shadowcraft, Warhammer, hell even GURPS had a tank and healer concept.

It's a very useful abstraction.

Your are right, these games supports roles, but not to this extends as current MMORPGs. Pen'n'Paper was a bad example, because it all depends on the DM. The player are able to choose whatever they want as long as the DM supports it.

A better example are single player/character games. Even class based games provides more or less equal chances to the player choice. It just changes the gameplay and the difficulty.

It is like searching for a proof of P = NP. In a community related game with balancing issues the designer should choose a good approximation of the balancing problem and never ever touch any community forum :-) Once the designer starts to listen to the community, the much feared nerf-spiral starts. (ok,ok, it's a burlesque, but the core idea is plain).

Quote:
**edit, I enjoyed your journal on random quest building. I am working on a similar system for random level generation except it's for outdoor worlds. It's still a basic graph like yours except I populate the nodes with pre-determined "stages" of content and of course I have to have some of them empty so the content distribution isn't regular. Good stuff.

Thx, I'm currently working at the story and quest-text generation which will be decribed in two additional parts. But I'm still testing the code.