Whilst designing the second and third tier of classes for Midmarch I came to realise that no matter how well I try to balance things certain combinations will ALWAYS be more effective. Sure I can do my best to even them out, but one or two will still dominate the rest no matter what happens.

Rigid class systems where you just choose say a Cleric, and you are always a Cleric don't have this problem so much since the developers know that there are pretty much only a fixed number of characters to deal with.

However in a skill based system, or a more customizable class system like Midmarch certain 'builds' will always rule supreme. My inital reaction to this would be to try and make sure that if there was an uber build, that it would only be effective under the right circumstances. So a PvP character would get torn apart against harder monsters, whilst a PvE character would do well against them, but never win a duel. However I'm ending up with combinations that would do well under all, or at least most circumstances.

In order to prevent this I started removing the cool and fun skills from my design, or nerfing them severely. So the new design is pretty bland, each character have only slight variations. Whilst they are still significent, they aren't exactly anything major.

So it seems I have a choice, have all the fun abilities and be prepared for Uber Builds and the player complaints that will go with them. Or have a set of combinations which are pretty bland but keep the balance.

It seems most MMOs favour the later (+3% damage against stone giants! etc).

One of the things I liked about Diablo II was the ability to customise my character, and to try and come up with effective new builds. I often ended up with characters that were completely useless, especially when I first started. However once in a while I'd come up with a build that was fairly powerful and I enjoyed beating people who though that a pally with a throwing dagger could never kill anything. However all the REALLY good builds (the ones effective at level 75+) were already worked out and everyone had them. It was sometimes painful to think that these cookie cutters would always wipe the floor with my carefully planned original build no matter what. There was never any chance to be the best AND original which annoyed me.

Can't seem to win either way. Give them real choices and they will end up being forced into the best builds and thus be all the same. Or give them fake choices with little impact and they will still end up being pretty similar.

Anyone have any thoughts or ideas?
If you want something that reflects real life, you'll have uber builds. Bill Gates, Arnold Schwartzneggar and Matt Bellamy are uber builds.

But this isn't real life and you have 100% control over your world. If you want balance, I don't see why -given that you are, in essence, God- you can't balance out the skills and combinations in your game in such a way that they'd be perfectly equal.

It's definitely not easy coming up with a balance (considering pretty much every commercial game with teams of scores of people haven't got it ...uh, oh wait.).

Okay so maybe it is kind of hard to get a balance.

Certainly a given skill may compliment another skill.
Consider two characters, both with skill level 10 in "Hunting"- the one with skill level 10 in "Trapping" too is going to be much better at survival in the wild than one with skill level 15 in "Word Processing" instead.

I guess you could make sure your game provides for all character types- if your game was focussed mainly on surviving in the Australian outback the player with trapping would do best, but if there was also a way for your word processing player to write scathing reviews of the Australian music scene (lololol what australian music scene) and bag lots of cash, then the trapper would no longer be such an 'Uber Build'.

Uh, hope you enjoyed my stream of conciousness.
There will always be best builds. Even with no builds, there will still be a best set of equipment, or best set of stat allocation; Failing that, there will be a 'best title' or 'best name'. Thats just how things work.

Removing fun factors to try and eliminate the amount of 'best builds' is a severe mistake, IMO.

Skill-based systems can be incredibly hard to balance, even if you're ridiculously careful with how you plan them out. There are some ways to prevent jack-of-all-trades builds, just as there are ways to prevent someone from super-over specializing.

There are many commercial games you can take a look at for suggestions here, one of which you mentioned.

Diablo II uses skill synergies to 'force' certain builds to be good/bad, despite the fact that they'd be awesome/[profanity] outside of synergies.

It also uses a resistance system, to prevent someone from completely focusing on one thing (Physical damage, one type of elemental damage, etc). It doesn't work as well as it should in a continual online environment, but its certainly something you can use as an example.

Its also a severe mistake to cripple builds to only be good at one thing (ie make 'PvP builds' and 'PvE builds'). Those will come about by themselves through min/max'ers, no need to push them even more.

I'd look into restricting builds based on skills being required by other skills, limited by stats, percentage-based skills (be careful there; Be *very* careful), exponentially difficult to reach 'top' skills and a variety of other methods that have been utilized.

Star Wars Galaxies, Diablo II, ShadowBane and UO might be good games to look at for reference; Even if you haven't played them, there is a wealth of information on the internet about how the respective systems work.
In response to Alathon (#2)
Examining my design currently I have a Brigand class which can only be obtained by starting as a Rogue. By so after going Rogue > Brigand you have the following skills.

Steal Gold
Steal Item
Dual Wield
Can't use heavy armour or shields.

Now I would cosider the stealing and hiding skills to be skills fun enough and useful enough to be worth having (hide is great for PvP and getting through trash mobs in PvE). Combat wise a Brigand is going to be sub par compared to a Barbarian, since their offensive capability even with two weapons is less and they lack armour.

Having played a previous game with a Thief character it seems that nobody really played them since you could never steal a worthwhile amount of gold, and assassins could hide aswell. Thieves were pretty weak. However recently that game has had a pwipe and now Thieves are probably the most played class since gold is valuable once again. But I can see people simply using these as 'mule' or secondary characters used to fund their real character later on which I don't like the idea of too much.

The above can be fixed if your economy works correctly (I think WoW's economy is pretty good). In comparison Diablo II's gold was almost completely worthless once you had been playing a while, only needed on the rare occasion you wanted to buy potions or something. I've tried my best to balance the economy in Midmarch but as a few people have played it's becomming clear that it isn't (some people have already managed to buy all the insanely expensive best shop bought stuff and have gold in the bank spare).

Anyway I'm rambling. My main point was that whilst this class would be attractive to some people who are prepared to be the underdog for fun skills, the majority of people are going to end up as a Fighter->Barbarian or a Mage->Sorcerer since these are the two that deal the most damage. Knights and Priests are great in parties but not many people like being a support character. Rangers and Adepts would be great at soloing but never really any good in groups or as PvPers except in certain circumstances. (rangers are good at fighting in forest areas for example).

I suppose the answer would be to give the lesser classes extra skills to make them more attractive, just some little sweeteners on the side (maybe a special pet or something). Things that won't really make them much more powerful, just more fun.

I suppose something along the line of elemental resistances and attacks would help with diversity. I may save this for later as more of an equipment based thing though. You could have the gear setup to make you partially resistant to all things or very resistant to one and weak in another. Of course just like Diablo the power players will have several sets of gear for each resistance.

I hadn't tried ShadowBane before but it appears they have the same basic idea of classes. In that you start as a Rogue, Fighter, Mage or Cleric and go on to the next class. The only difference is that Midmarch has a third tier of classes. I suppose playing ShadowBane for a while might give me some relevant ideas.... although once I start playing any MMO my work tends to suffer. Tough call!
My favorite example of a game where many builds are viable is Magic: The Gathering. Okay, maybe it's not a roleplaying game. But it is massively multiplayer, with hundreds of thousands of players worldwide -- and it is played online, by at least tens of thousands, if not more. (And it's sort of roleplaying, since you play the role of a powerful planeswalker ans spellcaster, but that's stretching it.)

The point is that in M:TG, there are hundreds of excellent possible decks that are played, and even established deck types change frequently as the metagame ebbs and swirls.

This is created by a few conditions:
First of all, there are five "formats" of gameplay and a player can play in as few or as many formats as he likes. The first is open format, which most players play without realizing it. There are not hard rules and no tournaments, and you play whatever cards you want.
Beyond that, there are tournament formats. These are Type 1 (all cards are allowed, but the number of certian powerful cards that you can have in your deck is restricted), Type 1.5 (same as T1, but the cards are banned instead of restricted. This is a huge difference.), Type 1.x "Extended" (as T1.5, but a different banned list and only new-ish cards are allowed), Type 2 (as T1.x, but even newer cards yet and another different banned cards list), and Block (only the very newest cards are allowed, yet another new banned cards list).

The banned and restricted cards lists keep the most powerful cards from dominating the scene. They try to balance them as they make them, but sometimes things fail -- when every build has to either include or make serious adjustments for a certain spell, they usually ban or restrict that spell once it becomes clear that the spells is hurting the fun of the game. Those spells are called "broken". Usually, the builds that use that spell are still very powerful even after the spell is banned/restricted, showing how "broken" the spell truly was.

On the other hand, the formats keep refactoring themselves. Even Type 1, the most permissive of all the formats rules-wise, is constantly changing because of new discoveries of deck archetypes, gaining in popularity of certain decks, and the introduction of new cards into the arena.

I am primarily a Type 1 player, and even in Type 1 there are at least 6 or 7 "Tier 1" builds - more than the number of good classes in most MMORPGs. Then there are probably a dozen popular "tier 2" decks, which are not super-powerful but are fun enough to play and win consistently enough to have widespread use in the game. Then there are decks like those I have been playing lately - "tier 3" decks which really aren't all that strong, but prey on the weaknesses of the tier 1 and 2 decks in order to upset the balance that they so precariously keep. So-called "rogue" or "metagame-based" decks keep the big boys on their toes - and because they are usually played only by their ingenuitive creators and a rare few others, the number of such decks is beyond easily counting.

Of course, there are always the crap decks that rarely if ever win, too. But people play them because they're fun, or when they do win they really piss the other player off.

Dunno if that helps. :-)
In response to PirateHead (#4)
I think the balance you are talking about here has been achieved with a shear mass of skills available. Diablo has about 30 skills per chracter and 6 characters making 180 total. Diablo II probably has about 7 good builds in 1.10 (last time I played). Magic has about 8 good builds (when not using special restrictions) as you say, but 1000s of cards to choose from.

The tournaments are interesting, but how would I apply that into my own game? Make a skill then ban anyone from using it? That doesn't seem fair or a good course of action in any sense.

If I were going to have an ingame PvP tournament I could do this to keep it a little more balanced but for the most part restriction would be a bad idea.
In response to Midmarch (#3)
When a game centers around combat, classes/skills that are optimal in combat will reign supreme. Just the way it is. If you don't want to focus on non-combat activities, then you need to allow different routes in combat that make all the classes equivalent or useful in combat. Some primary roles:

Damage dealer- This is a class that excels in raw damage output or circumventing defense. Ideally, this role is not also a primary tank, though that is a good secondary role. High usefulness in solo play. Recommended concepts: Barbarian (high damage output, limited armor, evasion), Rogue (High accuracy, critical hits, special defense evasion IE armor penetration or similar, low resilience/armor), Fencer/Duelist (similar to rogue), Wizard (High damage spells, slow recovery, poor defense)

Tank- This is a role that can take alot of hits. This role is often the lynchpin of grouping and usually requires good support to stay alive/minimize damage and to make sure bad guys stay focused on the tank. High usefulness in solo play. Concepts: Knight/Fighter (High armor, low evasion, decent damage output), Cleric (combo of support and tank, rather poor accuracy and evasion), Monk (Good evasion and resilience, but low armor)

Buffer/Support- Usually a spellcaster, this role is very useful for increasing ally abilities such as evasion/defense, armor, accuracy, etc. allowing a group to take on more powerful threats. Generally moderate fighting and defensive abilities are recommended so as not to eclipse pure damage and tanking classes. Moderate usefulness in solo play. Concepts: Shaman, Priest, Bard, Enchanter

Healer- Also generally a spellcaster, this role is valued for its ability to restore stamina, health, magic etc. and is crucial for quick recovery as they generally have solutions to status ailments. Again, lessened fighting abilities are recommended to compensate for the increased resilience from healing. Largely the inverese of the damage dealer. Moderate usefulness in solo play. Concepts: Priest, Light Wizard, Paladin

Debuffer- The inverse of the Buffer, Debuffers generally lessen the abilities of opponent's through status ailments and effects. A huge force modifier, combined with a buffer, they can make damage dealers incredibly dangerous. Debuffers offensive ability tends to be limited to damage over time and may have weakened defensive abilities (in which case pets are recommended to supplement). Moderate usefulness in solo play. Concepts: Shaman, necromancer, evil priest, dark wizard

Manager/Crowd Control- An essential role in grouping if there is any kind of AI, the manager makes sure only desired enemies are engaged, that they stay engaged, and to the right targets, usually through taunts, magic, etc. This role often has abilities that limit or modify opponent aggression (or redirect it to other targets!). Moderate usefulness in solo play. Concepts: Bard, enchanter, jester/dancer

By splitting out roles, you help ensure a greatter number of choices and roles. Optimization is unavoidable and balance is tricky if you don't stick to a zero sum formula, but at least this way you have 6+ optimal configs instead of one or two :-)
Well, a design that seems to effectively avoid this situation is focus on more then one thing. The game could be a real-time combat game, a strategy game in which you command a group of mighty warriors(players, AIs), or an intelligent game in which you are a wealth-seeking trader traveling around the world. Going by that build, a mighty warrior would most-likely not have enough time to investigate in bargain skills, and a wealthy trader wouldn't make a very successful general in the royal army. Going by such build, you could never have the ultimate build, as one subject could not best at another. Going by the same system, you could also add, instead of skills as suggested by Elation, classes that compliment the other. A mighty warrior could make quite a good general: striking fear into the hearts of those who oppose him; giving confidence to his comrades.
In response to Northern Rabbit (#7)
Merchant characters will just be used to fund their REAL character (combat specialist). That's not such a bad thing, but I think having a Thief type class is better. Same kind of thing but much more desirable.

Trade is something that is only much good between players. Having all items buyable in a shop is plain boring so the need for a merchant class is much less.

In response to Jmurph (#6)
This is pretty much how I've tried to design the roles. However due to it not being set classes it makes these roles harder to create. You choose something like Fighter > Knight > Inquisitor, or Cleric > Priest > Inquisitor etc. So there are several different versions of the end class. I create all the skills then see what each different combination would be (I think there are about 50 possible combinations). I keep finding a few that are obviously very powerful due to synergy of two skills. As an example I have situations where it's possible to accel it multiple roles. I also have combinations that don't compliment each other at all, leaving a heavily underpowered character.

I am getting there, but it's not easy to keep each character type fun, balanced and different enough from the other classes.
In response to Midmarch (#8)
Midmarch wrote:
Merchant characters will just be used to fund their REAL character (combat specialist). That's not such a bad thing, but I think having a Thief type class is better. Same kind of thing but much more desirable.

Trade is something that is only much good between players. Having all items buyable in a shop is plain boring so the need for a merchant class is much less.

Perhaps I should clarify: I wasn't quite talking about these classes, but the possibility of adding different ways to go throughout the game, and thus playing differently. In theory, that should cancel(or at least make it harder) the possibility of uber builds, as a brute warrior could never make a successful scholary.