Correct me if I'm wrong, but the third chain ability sounds much like a crit of some sort, and you're suggesting that the critical chance be obscured and tweaked on the fly by developers in order to balance out classes.

Having played a lot of TF2, where the developers have done a fair amount of fiddling with crit % rates, I can definitely tell how often I get crits, and when the probability has changed.

It sounds to me like the main question being presented here is how much of the game system should be presented overtly to the player, and how much should be handled covertly by the developer. Tweaking covert systems on the fly could indeed be a way to go, but this could very well piss off those who want to know precisely the best min/max arrangements. For example, if I were choosing a skill in a skill tree and I know that skill has a great critical version(for example, aoe does more damage and stuns instead of slows), it would be very important to know how often I would get that critical version. Tweaking something like this behind the scenes could prove to be very frustrating to the player.

I'm not really sure if there's a good way to introduce a sort of buffer between developer and player when balancing skills and classes, and the best proven way so far has been statistical analysis combined with listening to the community.

I think there was another blog posting earlier on this year about how randomness is a bad way to balance a game because players hold unfavorable random events as unfair. If you're relying on the randomness of "third chain" becoming available, and one player gets it every time and the other player never gets it, does this seem like a fair balance, even if, overall, the odds of getting the third chain are even over a multitude of fights?

@Jesse: The 3rd chain ability isn't a crit. Any weapon/spell hit or ability can crit. Also, victory isn't purely dependent on whether the 3rd chain ability is triggered, but it helps. Crits are tuneable with good gear and are more predictable. The way the 3rd chain works in practice in any given engagement sometimes it will trigger only once, sometimes twice and often not at all. It is random. One could keep track of those to figure out a percentage but it would take a few hours of combat to gauge what the percentage is with any reasonable margin for error. To make it even harder to gauge is the fact that there may be other weights that impact the roll. Based on my own observations I'm guessing that the ability is more likely to trigger when fighting weaker opponents. Given an extended PVP battle with a variety of character levels, it would be hard to isolate what the class weight is.

@Simon: "... players hold unfavorable random events as unfair." I don't buy it. It's a game for crissake. Play D&D and someone is going to roll a 20 or a 1. Yeah it sucks if it's the monster attacking you, but you have the same opportunity to roll well against the monster. How fun would it be if by judging by pure statistics, the player with better gear and maybe one level above another player should win every time? The act of NOT being sure, and it NOT being competely deterministic adds an important element of risk that raises the heartrate and makes the game exciting. Personally, I wouldn't play a game if victory were purely deterministic. There has to be some randomness. Would anyone bother to watch the Super Bowl if there was NO chance that the underdog would win? No, randomness is important.

Timothy, thanks for raising an interesting issue.

The pitfall I see in trying to add too much unpredictability is that you end up having to design encounters that are tuned for greater variance. This means you'll either have to make the encounter easy enough that you can defeat it with a string of very bad luck, or have an encounter that is impossible no matter what the player does if their luck is bad (and to me, thats not a game its a swindle.)

But lets say you take the first approach and ensure that even with a string of very bad luck you are still able to defeat the encounter. Now, what happens when you have a string of normal, or even good luck? The encounter becomes almost laughably easy.

If you visit some of WoWs harder core forums (try damage dealers) you'll often see complaints about the weight given to RNG [random number generation] in encounters. When players feel like their skill is superfluous to their success they feel like the buttons they push dont matter and the competitive feel is lost.

Add to that the fact that tuning these encounters makes for encounters that are either impossible or too easy and you find a very narrow window for the inclusion of randomness.

Now, lets take a look at something you tossed out there at the beginning of your article -

"I would NOT do what WoW did and start blurring the lines between character classes by giving them all similar but renamed abilities."

In my opinion, I think this is the most promising lead for designers looking fore a way to eliminate imbalance.

Back in the early days of WoW and 40 player raids, every class had its role. end of story. Alt specs were not really viable, and every class was necessary. Personally, I played a hunter in WoW (I was a damned good one!) which is primarily a damage dealer. Back then I didnt care in the slightest that rogues locked up the top spots on the damage meters because I loved my role - I got this thrilling and insane challenge of trying to "pull" all of the mobs to the spot where my group would do its fighting because of my special talents as a hunter. There were also encounters that required me to use a special tranquilizing ability on monsters to keep them from enraging and mowing down our whole group in very short order. This made hunters necessary even though their primary role as damage dealer was arguably better done by another class.

In that way we all had a place and felt valuable to our raid. I didnt care that I would never catch a rogue on the damage meters because I had a vital role, and mages never cared that they couldnt catch me because they were too busy turning things in to sheep.

In modern WoW, post generalizing of the classes, players are livid when their class dips below others on the damage or healing meters because they have no specialized role. Because all classes do the same thing they have no value beyond being the best damage dealer or healer, and if you're not number one, why are you even here?

Given this example I think its a valid solution to give each role a niche that makes them wanted and necessary parts of any group. Players may compare themselves against other players of their own class to get an idea of their own skill level, but they dont need to worry about whether or not they are the best of all the classes.

Now, when you talk about PvP you are taking on something beyond the scope of my solution. This is a far more complex scenario to balance depending on your expectations. i.e. should any class have the chance to beat another class in a 1v1 fight? now you have to solve for a host of balancing issues, but basically it works out to dmg out, dmg in. mitigation, healing, 'crowd controlling' abilities, range and distance closure must all be tightly tuned to give each class a chance at winning.

@ Timothy

It depends on whether you are trying to make a game of skill or a game of chance.

Personally, I have no interest in a game that doesnt take skill. Further, I have more interest in games that present enough of a challenge that skill and aptitude are required to be successful. Further still, I want skill to be such a factor that I can some times beat people with better equipment simply because I have more skill. that is a sign of a great game to me.

@Richard Putney: I agree that PVP vs PVE has entirely different goals and as such would have different measures for determining what sort of buff or nerf needs to be applied. While the debate over randomness will go on forever, I think there is a consensus that there will always be some sort of randomness as long as statistically over many encounters the results are predictable and you win/survive an expected percentage of the time and there are things you can do to groom your character to improve the percentage. The effort to make MMO's a PvP platform are what's behind the generifying/flattening of player abilities, like a traditional shooter where everyone starts with the same amount of health and the equal chance of finding the BFG and owning the match. Players in such PVP contests want their individual skill to shine through and they are less interested in role playing.

@Richard: I think we are in agreement that skill (player choices that benefit from experience), stats (for grooming/improving your character), and randomness (uncertainty) are all important in making a game fun. IMO, good games find a way to factor in all three.

Apart from what has already been said, if you're using static numbers inside the black box, players can and will parse the data and find out the approximate number. Thus the tweaking of values would eventually be visible by those people who are most likely to care about such values.

If you've played WoW, maybe you've heard of 'theorycrafting'.

Abilities that are directly affected by a player's stats (eg gear/race/faction) will never EVER be balanced without sacrificing uniquity amongst players.

Here's what I think:

A) Throw out gear stats and just add a bunch of options to tweak the looks of gear so that the player always feels and looks exactly the way he/she wants to look.

B) Throw out most of those useless and over complicating stats. Base stats on the natural laws of biomechanics and the associated body type. EG endomorph, ectomorph, mesomorph

C) Come up with a real world combat system that players can better relate to and understand. I mean if you were to break it all down, you're either DEFENSIVE or OFFENSIVE when you are in combat. Beneath those 2 main factors comes 4 underlining sub factors: striking, grappling, evasion, and deflection. Based upon those factors, develop and design a stylish art form for each type of weapon in the game and then mocap it.

D) Based on the above, create a combat system that is PHYSICALLY SKILL BASED and throw experience and leveling out the window forever.

There's a reason why there isn't that many good action packed mmos (most are more like cheesy social networks than alternate realities) and I think it's time for the designers to finally put an end to these ridiculous mechanics that need constant tuning and finally focus on the technologies to fully immerse players physically (as in Wii/AVATAR physically) within these virtual realms. For god's sake, most of those who play mmos could really use the exercise anyways, so you'd actually be doing something good for humanities health for once rather than just sucking them dry of all their free time.

All in all, the reason why there is so many balancing issues is because players are being limited by these stats based systems. A better approach can be taken by utilizing an ever evolving combat system that is physically driven by the player and not by the numbers. You open up the world of unlimited possibilities to the player if you spend the time to create a true to life combat system. The best example to go by is an actual martial art since since many of them have been in development for thousands of years.

A typical form in martial arts consists of 12-18 moves and most of these moves are a series of combos consisting of multiple types of actions: striking, grappling, evading, deflecting. The first thing you need to figure out before developing an art form is to determine whether or not it will be multidirectional (more than one opponent) or a one-on-one type combat system such as boxing. This is all dependent on the stance which is either defensive or offensive. Although depending on how wide or how close the stance is, also determines how many opponents one can take on. For example, a boxer always keeps his fists in front of him because the time it takes to react to an opponent is reduced by the simple fact that there's a shorter distance between his fist and the attackers face. Although in a situation where there's more than one a opponent, a wider stance is needed, in which case you'd place one arm in front of your other arm behind you.

And this is where mmos are going wrong, there not making their combat systems physical dynamic (in terms of players having the physical freedom to react to every type of situation) enough and so the balancing issues will always be a problem.

Here's a simple diagram of every two-step combo move one can produce: http://farm4.static.flickr.com/3192/2670781843_96b03a129e_o.jpg

Well, it's not based on every direction, but it would allow one to develop an infinite amount of forms which could also translate into special abilities within games.

...and that's my 2 cents

Before you can talk about "how" to balance, you need to define your goals.

Does "Balance" mean that all characters are equally capable of dealing with all situations?
Does "Balance" mean that each character is equally important in a group but not necessarily equal on their own?
Does "Balance" mean that different characters will be better in different encounters but the sum effect of all encounters in the game will make the characters roughly equal?

There are plenty more questions to ask, and these are just for PvE. Richard's post is mostly in agreement with my line of thought, except that I see that more as a fundamental design decision for a completely different game experience than a simple balance technique.

Likewise, you can look at the discussion here on luck. Before you talk about the best way to balance, you should talk about how much luck should be a part of your game or what it can accomplish. If you take a pre-existing game design, then add an element of luck to try achieving balance, I feel like you're designing backward. On the other hand, if a strong luck element is already part of the game, tweaking it would be a good approach to improving balance.

@ Timothy
Even in a game of DnD the player that rolls 20s often gets looked at darkly by the rest of the team, while the player who always rolls 1s gets quite frustrated. Now when talking randomness, players don't mind as long as it's not a negative random chance, for instance, missing causes more annoyance than an opponent dodging, we don't mind an opponent critting against us, but we hate it if something we do affects how much damage we take (aka rolling a 1). Additionally if it's something that can be planned for like spell damage ranges 20-200 etc, players don't mind those sorts of random. But arbitrary "this skill may become available" style affects tend to annoy people.

I always preferred the use of a metagame to balance pvp elements. In a game like wow this is next to impossible due to the high reliance on gear and the fact that players take all their skills into battle with them. Games like MW2 and Guild Wars though, limit your selection of items or equipment. So while one spell or gun proves to be the most powerful, as long as there is a selection that counters it, players can take the sub-optimal skill that protects them from that ability. Having played Guild wars a lot this became quite evident. For instance, builds focused a lot around conditions to deal with powerful direct healing, so healer builds changed to sub par skills to deal with conditions, other builds then switched from conditions to heavy interrupt focused builds etc etc.

It lets the players balance themselves, it doesn't even matter if the options are the most optimal, as long as there is something that's a solution to any potential problem, players will find it and make it work.

@ iLLin CruX
'Grats, I think you just invented street fighter II.

@ Timothy
/Agree

Lets try something really different as far as introducing 'randomness.'

Lets say our game takes place in a magical and highly unstable world. The five primordial energies of the world are constantly swirling through the existence, waxing and waning in their strength.

Fire, water, earth, air, and spirit are constantly in flux, and at varying intensities in the players vicinity. Now, the class, elemental attributes of his gear, and the spells he casts are all affected by these constantly shifting forces. If he chooses a moment to cast his frost spell when water energies are very low, he may have diminished slowing effects and damage. If he casts this spell during a moment of peak water energy his spell may gain a critical effect. Likewise, peaks (or lulls) in various energies could proc gear or class abilities. This way there is variance in the mechanics, randomness and unpredictability but it can also enhance the need for skillful play.

@Kumar

/agree. players will eventually figure out the proc per minute rate and will be able to estimate their effectiveness the same way the developer ensures that this ability is balanced. Well, if your players get obsessive about it as they do with many MMOs they will. I suppose thats a good problem to have.

@Richard I like your idea of fluctuating buffs, but I usually associate that with the metagame in order to shift the tide of battle and allow lowbie's the chance to maybe make an assault and recapture territory. Aion does this with the artifacts in the Abyss though players activate them (it's not random). In any event, if players are used to their ability varying a little by the phase of the moon or position of the stars or what have you, they might not notice a hidden buff or nerf for their class or the fact that their particular planet of influence is gaining strength or waining.

Randomness is the opposite of balance. Randomness is what a designer does when they've given up trying to balance something. =P

As to hiding the numbers... Kumar is right, players will parse and figure it out, and woe to you when they find out a huge factor is a big random number. Making it so sometimes the end result is that the person who should lose will sometimes win due to a random factor is a cop out. A good designer will analyze the situation, find out why (or even if) it's a balance issue, and make a minor change to fix the issue. The more minor, the better.

Sometimes designers make the mistake of making sweeping changes on both sides of a balance issue in order to show players they're doing something. RESIST THIS!

The key is one or two very minor changes per balance issue because the more dramatic the change, the bigger the chance of further balance issues coming in from other factors (great, we balanced warrior versus paladin, but now paladins rape all, fail).

Ok, what would I do?
It's something I brainstormed many years ago, during that post-MMO-discovering "I wanna make an MMO" phase. I won't fill every gap or take into account every possibility, so it's just a starting point.

-----

First, there are no levels, or at least not in the same sense most of them have. Maybe even not classes.
The more you do/see some ability, the more you improve that one ability. You can learn new abilities by watching them (if your "watch-an-learn" ability is good enough for it) or by being taught by other players or NPCs.

So into the "balance": every ability has two arrays of values: offense(with) and defense(against).
- Offense (when it's the case) is the char proeficiency when performing the ability. The more tha char perfoms such ability, the better he/she becomes at it. (ex.: throwing an arrow with a bow);
- Defense (when applicable) is the char experience/ability to defend/avoid the effects of such ability when another character used it against him/her. The more the character faces such situation, the better he/she becomes against it. (ex.: defending against an arrow thrown by a bow).

It makes a mix of demand/value rules (in which you "nerf" popular abilities and "buff" rare ones automatically) and some kind of "hidden trick" mechanics. So the devs can introduce more and rare abilities over and over, even slight tunes in some to make different version to mantain that "ecosystem".

-----

Anyway, it's just a start for other possibilities, and every solution will have it's pros and cons.

I think the article Simon is referring to may be this one by Soren Johnson:
http://www.gamasutra.com/view/news/25127/Analysis_Game_AI__Our_Cheatin_Hearts.ph
p

I think what he says is true - when players get a long string of unfavourable outcomes, they interpret it as the game cheating, even if rationally they should know otherwise. In fact, even when players *do* know otherwise they often become frustrated; anyone who's ever seemed to have an impossible time getting a piece of loot to drop in an MMO knows that frustration well.

That's one of the main reasons that a GM in a pen and paper RPG has so much control - so that the randomness of the die rolls doesn't take over. When I used to GM Shadowrun, for example, if a player got a string of critical failures I would turn it into a humorous moment if the situation allowed, or if not, create a great epic moment for the rest of the team to come together and save them. But without that GM in place, those random rolls just turn into frustrating player death and loss of progress, and no one wants to lose a bunch of progress because of randomness.

I don't trust randomness to bring anything to the table, just look at the drop tables accumulated for every single monster and area in WoW. It's all "hidden" information but the players have defined the percentages up to 3 or 4 significant digits.

I think if the devs keep good in-game metrics and have a strong community they can balance stuff quite well, or well enough.

@Dave and all those against randomness: If the single most determining factor in a battle is randomness, I'd agree with you, it doesn't make sense. It would piss everyone off. Of course skill and your gear/stats grooming efforts should influence your success, but they shouldn't guarantee success 100% of the time. It's a weighted system, where some randomness plays a part. RPGs (and I'm focused on MMORPG's here) have always had dice rolling. Does level 10 player always kill a level 1 player? Of course, despite the random die rolls, even a few perhaps slanted in one's favor wouldn't change the overall result. But a level 9 player might kill a level 10 player because some randomness in the combat and of course because of skill. The random factor handles the edge cases: it doesn't perform miracles that might piss people off or make them feel cheated.

If you play D&D, does your DM always role the dice in front of you or does he often roll the dice behind the dungeon master trifold screen? The DM orchestrates the experience he wants maybe in order not crit on you or similarly steer the game experience in a more favorable direction. That's what the black box solution I suggested is doing - some rolls (NOT ALL ROLLS) would not be seen in order to balance out the abilities and classes over the course of many encounters.

I would venture to say that players who don't like randomness at all are probably PVP focused and don't have a role-playing background and should probably stick to MP shooters (or spell-shooter).

Re: Game Parsing: Who says your game should provide an interface that would allow the parsers to capture anything? WoW does and it's convenient but it's also dangerous and helps players to cheat and twink the system. It's like counting cards in Vegas, you just don't allow players to do it or make it really hard for them to do it by using a four-deck shoe.

@Timothy a fair degree of randomness (which personally I'd place around a 40%) in mechanics is great. I totally agree that a battle should never be won before it begins (due to gear/level).

However I disagree with the idea of hiding the percentages of abilities and such from players. One of the things I always appreciated with D&D was that everyone knew the chances of events and played their odds knowing well in advance their chances of success. It's clearly stated in the books and worst case the DM will state upfront what are your chances of performing some arbitrary action before you roll. It makes it more exciting that way (IMHO).

In almost all cases, and areas, transparency is better than black boxing.

Balance is called balance for a reason, it has to be equal on both side, like a real balance. If you randomly throw rocks at the balance, some will land on the right plate, some on the left one, and some will just miss and fall on the ground. People will instantly see there's not the same amount of rocks in each plates, and then they'll also figure out how many are just not on the balance.
Now let's say someone is in charge of putting ALL the rocks on the balance and it has to actually be balanced. These rocks are all different, some are heavier, some are bigger or smaller, you cant put all the big rocks on 1 plate, it won't fit even if it's balanced. You'll need big, small, heavy, light rocks on both sides to make as many people happy as possible.
If we translate this in MMO balance mechanic, all your types of rocks represent different skills of only ONE character class and they all NEED to have an equal chance to balance versus ANY OTHER.
If you think this is impossible, you are right. Unless it's 2 same class with same gear same lvl same skill build, it will never be balanced. In order to """"fix"""" this problem most MMO have a "rock, paper, scisor" balance system, where one class X beats class Y, but class Y beats class Z, who can beat class X. It's a never ending cycle or boredom where you require specific classes to do the job your own class can't do. Now to someone who just looks at it and never tries it for real, this seems like a pretty decent balance for teamplay, where you "theoricaly" have people with you who can do their job and help you while you do your job and help them. In reality this does not happen, MMO players will find the turbo combo of class that can disable their oponent's game or nuke out their key member in the first second. In a game where people "blindly" choose their character class (by that I mean they read the description but we all know too well class descriptions are everything but accurate) and level up for several hours/days/weeks and want to play with friends, this may not work, their chances to win are reduced if they dont have the right combo, anyone who played arena in WoW knows what this means.
So basicaly if each classes dont have equal chances against any other class, teamplay will not be balanced either, it only hides the most obvious unbalanced but it doesn't solve any problems. Different classes cant be balanced against every other, this is humanly impossible, and having only 1 class is not a MMO, that's a shooter with no special ability involved, so the only way to win is to just play better than your oponent instead of having the most stun or bigger random crits.

To get back on the "random chance to trigger" idea, mathematicaly if each side has the same % chance of triggering the same kind of skill, it is balance, but in practice, this will let one player use it more than the other during one fight, and this is not balanced. Players will never see balance if they don't control what their character does, the loser will call the winner "lucky" and if they fight again, the loser may become the winner if he gets lucky. Now you may think it's balanced because both players have 1 win each, but in fact if they do 10 more fights and one player wins 8 out of pure luck even if his oponent played better but his unlucky numbers wont let him win, there's just obviously no balance. Random is not balance.
Situational abilities like the "ice spell buff in a high water energy area" thing can only be called balanced if both players have the same oportunity. They both have to be in a high elemental energy area and use the correct spell, lets say the area are about 10 meters wide, they can be in different elemental areas, one in water, one in fire, they both use the right spell, all is fine. But if one is in an area and the other is not, there you go, unbalanced.
MMO players need to have the same amount of options as their oponent in any given situation or they will see unbalance.
Any kind of disabling ability gets the game further away from balance, one class can silence the caster, caster can't do anything, he dies, he cries to the devs, devs give the caster a new defensive skill to survive the silence time, which has now been nerfed by the devs, now the caster has it too easy and more people start playing caster, etc you see where I'm going with this. The very big majority of MMO players don't care about their class role, they only want to win, giving every class a specific job doesn't do any good in a player's point of view as all they will see and remember is everytime they lost all their fun against their counter class.

But to put it in one line I would have to quote what Robert said above : "I think if the devs keep good in-game metrics and have a strong community they can balance stuff quite well, or well enough."
Keyword being "well enough", MMOs will never be balanced, no matter what, but they can come close. We can have a "this is the most balanced MMO out there" but we will never see a "this MMO is balanced" game.

On the randomness topic, here is an interesting article on player perception and expectations of it, and how it has and is being used in the Magic cardgame.

http://www.wizards.com/Magic/magazine/Article.aspx?x=mtg/daily/mm/37

Basically, their players despise randomness upon which they have no degree of control, or that they don't expect to find on the medium.

@ Timothy
In my example of environmental randomness you're introducing a world and a combat system, so there would necessarily be some narrative to explain how the world works and how it affects your character. The rest of your concern is addressed through good UI.

on another note, it doesnt matter whether you give people access to your combat mechanics numbers. If there is any consistency at all to your world they will find all thresholds and proc rates.

@ Michael
I agree with your sentiments on randomness. In my example I think I failed to make clear the rate at which these energy fluctuations are occurring. The energetic levels in your proximity could be very different after five seconds of combat. This ensures that there will be peaks of multiple energy types within the span of a single encounter, and the flux cycle would be balanced around the duration of a single fight for standard areas.

Now, given this system a player could hypothetically have fire abilities and frost armor which allows him to mitigate unlucky periods for his offensive abilities with armor that is powerful when his offense is weak. He may also choose to stack his offensive abilities (esp. in the case of group PvE.)


@ All
I agree with the general consensus that randomness =/= balance. To establish balance in a complex combat system with multiple roles and classes (or in the example of classless systems, multiple specializations) you need to know alot about your other design objectives for the game. I agree that the only way to generalize balance without that knowledge is sameness, and this defeats your objective of distinguishing multiple roles, classes, or specs.

I haven't read all the comments here but I thought I'd chime in by pointing out something which Damion Schubert of the SW:TOR MMO team said in a talk, and which Raph Koster liveblogged here:

http://www.raphkoster.com/2006/09/08/agc-damion-schubert-moving-beyond-men-in-ti
ghts/

Basically that, at least in PvP, having at least some degree of tactical transparency (i.e. know exactly the abilities and numbers of your enemies) is better. I think this could be extended to saying that players are better of knowing the numbers behind their abilities and their enemies'...

@Richard
Now I understand your idea, and I think it's good. It can explain the variation in damage or mitigation with a logical system instead of dice roll. This could also add a lot to the gameplay if people have something to tell them aproximately what kind of element is around them or what is coming. If the player pays close attention he can control the battle by using the right spells at the right time but during an intense fight he can quickly lose track of it and use the wrong spells.

@Michael
Thanks! Yes, thats the ticket. I'm envisioning a bar graph style indicator where you can see the rising and falling energy levels and respond to them. the rate of increase and decrease could be predictable while the direction (higher or lower) would change at random or pseudo random intervals.

In addition to your observations the player would probably be quite concerned with the energies that are empowering his opponent. This could be a critical aspect of boss encounters.

Another thing I like about this concept is the amount of player engagement and reaction required as opposed to repeating the same combat rotations ad nauseum.

Class designers act too much like a government after the games released. That's exactly what players don't need. Instead spend time making new zones and new interactions.

Your warrior has three ways to heal himself? You mean like enraged regeneration, which only heals 30% of your health on a 3-minute cooldown? Or last stand, which requires you to spec protection, and only temporarily heals 30% of your health on another 3-minute cooldown? Or do you mean like... actually, I have no idea what the third way is. You're not including a blood draining enchant or trinket, are you?

WoW has indeed homogenized the classes a bit in the last expansion in the interest of increased raiding accessibility (well worth the trade-off, in my opinion), but there's still plenty of distinction between warriors and paladins, even with regard to warriors' (extremely limited) self-healing abilities. (Seriously, a blood death knight would have been a much better example here.)

Anyway, regarding the question posed, the first step to MMO balancing is establishing what exactly you're trying to balance. By far the most unbalanced part of WoW is various 1vs1 class match-ups, but Blizzard has explicitly said that balancing 1vs1 encounters is not one of their design goals. It is, however, perhaps the most balanced MMO in the history of gaming with regard to pve raiding -- all 30 class specs are perfectly viable for any current encounter, even if some slightly (and often temporarily) outshine others.

Balancing pvp is far more difficult, and in this case I don't think any amount of theorizing will help you accomplish anything. You just have to playtest things and see what's broken. When WoW was first released, fears and stuns had no diminishing effects, which just seems laughable now. In the case of a game like Guild Wars, you just have to look at which skills are being used incredibly often and what has been deemed mandatory by the players -- those should be the first to get nerfed.

Balancing pve and pvp with the same skill set is so close to impossible that I wish MMO designers should just stop attempting it. In the first expansion we saw WoW banning high-cooldown abilities from ranked pvp matches (arenas), and just recently we saw a universal healing debuff applied to the pvp areas in the game. Why this took so long, I don't know, but the game has a long history of nerfing abilities out of pvp concerns when they were functioning just fine in pve, and the issue has built up a lot of resentment in the community with people who only care about one aspect of the game and not the other.

Ultimately, MMO balance is purely about preventing people from being angry. They're angry if they feel like their class is useless, and they're angry if another class is dominating them. That should be the focus of the balancing process. Never let a class or spec feel dead/useless, and never let one ruin fun for other people (this isn't exclusive to pvp, either -- it's also not much fun when you feel obligated to reroll to the dominating class in order to feel viable in pve).

Before I begin I'd like to point out that I am woefully ignorant of the complexities of WoW mechanics; I know the basics but not much more than that.

What I'd like to point out about MMO classes/balance is this: all classes need to be created equal so that no classes will be avoided. There will be exceptions to this, but for the most part once people learn the game they will choose the better class eventually.

With that idea (probably horrible I know) down, perhaps game designers need to place more effort into creating a system of greater chances in battle? Reality proves that most of life is pure chance. So wouldn't it make it more of an unequalizer to have a great chance for random occurrences? I agree with a statement above that this will make players angry when chance sides against them. So I'm open to a logic explanation to my question.

In addition, what does WoW do for random object placement? As an idea, couldn't a powerful object be placed in a location, only to be used by the bearer while he/she has it? This would create a situation of other people coveting the object which will bring a more personal aspect to the game play. Whole battles could be fought between groups to obtain an especially powerful spell book/sword/cloak which would create a story line among the players, instead of needing it to be written into the game by the developer.

These are just questions/and ideas from someone who doesn't play many MMOs, so forgive my ignorance if it shows itself.

Cheers!

@Richard Putney

No I did not just invent street fighter 2 (i believe the art of physical combat has been around much longer than that game) with what I said and in fact, I don't think I made myself clear enough. So let me clarify, if those 4 factors (striking, grappling, evading, deflection) are not within the physical control of the player, than there will never be balance. A player needs to be able to control them so that no class ever becomes over powered. Even the weakest hitting player could take down the strongest if one were able to physically control them (with heavy use of evasion and deflection tactics of course)...the only limit would be their reaction times to effectively executing maneuvers at the right moments.

It's ridiculous to try to quantify every aspect of combat using numbers without first building a system that allows the player to physically control the very foundations of it.

3D interactive video games are NOT tabletop games and they never will be...So lets stop assuming that everyone is a DnD player from the 70s/80s where nearly every outcome is determined at the role of a dice.

Here's a bit more clarification on what I think players need to physically control, but let me break it down once again:

The two main factors of combat:

Offensive
Defensive

The sub-factors:

Striking (Offensive)
Grappling (Offensive)
Evading (Defensive)
Deflection (Defensive)

Manoeuvres:

Striking: meleeing, kicks, punches, elbows, kneeing, ranged attacks (anything that deals direct damage)
Grappling: throws and joint locks (stuns, bashes, rooting etc in the rpg world...anything that restrains an opponent)
Evading: tumbling, dodges, jumping, ducking (anything that allows the character to completely avoid an attack)
Deflecting: parries and blocks (reduces the amount of damage taken from an unavoidable attack. Although on a side note, experienced fighters will learn to also deal damage with the use of effective parries and blocks.... I've seen people who have actually had their bones broken from a very hard hitting parry (which typically happens between a cocky experienced fighter and a beginner who lacks the 'Bone Conditioning' to with stand the force) and I can't count how many weapons I've seen broken from precision blocks and parries as well...)

As far as ranged goes, this non-sense with heat-seeking arrows and spells needs to be dropped once and for all and be replaced with any type of shooter combat system that requires one to actually AIM! Otherwise, all of the above will be defeated. And If you're going to add stats to players and gear, they should consist of things that the player can't possibly physically control within the virtual realm and is obtained over the life time of a warrior. Such as the speed and force of an attack, resistances to things that are both physical (bone conditioning) and magical, et cetera.

All in all, this is coming from someone who was a competitive fighter for 8 years (up until 5 years ago) and maybe I don't know all of this game design lingo. But I do have extensive knowledge in hand to hand combat and with that said, please believe me when I say, you need to allow more physical control over specific elements if you want to achieve the balance you're looking for while also creating uniqueness amongst all classes and players. I mean, you could have a class that can throw lightning bolts that deal instant death, but so long as one is able to physically utilize any number of evasive manoeuvres to completely avoid it, well then you can say good bye to balancing classes and focus on more important things...like adding quality content.

In closing, try to think of combat as a spoken language between opponents which consists of only 2 letters: [O]ffensive and [D]efensive...so it's binary. It's very simple to learn how to speak it, although it is very hard to read, and that's the way it should be at a beginners stand point. But as you gain experience, you begin to learn what your opponents are saying and by studying the patterns in their actions (some are harder to read than others), you begin the path of mastering the secret language of Anticipation and Deception.

Game Designers don't want classes to be balanced, players want to feel that their choices were justified.

To expand on that: Players want to feel like there were several good choices, and that they made a good selection from that menu. They complain about balance issues when they feel, rightly or wrongly, that one class (or weapon type or race or faction or way to spend skill points) is clearly a better choice. Game designers make classes to fulfill some in game purpose, normally roles such as tank/healer/dps or some variation of that. There is no way to mathematically balance "Can wear leather armor" with "Reputation increases 10% faster" or "Can quickly lose all threat once every three minutes". Game makers don't want that.

Game designers of MMOs with active paid subscriptions do want to keep the player based engaged, so do expect new patches that change how the game is played, and player's perception of what classes are unbalanced. They will employ Community Managers who will funnel info from the players to the designers ("Guardians are overpowered") and from the designers to the players ("Joe plays the Robowarrior class himself, and he says it owns.")