I have a good article about game balance (author - Zileas, gosu StarCraft player, designer of StrifeShadow RTS, site - shambler.net). It may help. At least it's best i ever see. It seems that Blizzard used some of that advices.
LONG POST
Lecture 8: Play Balance
Game balance, in the broadest sense, is derived from two major axioms of game design. The first is that players should have options and choices -- a game without choices is usually not really a game... choices and the interaction involved are what elevate a game from reciting a script, or watching a movie.
The second is that games should not have unnecessary components that dilute the gameplay. Choices that are never a "right" choice, or are almost always a wrong choice are not really choices at all, and are just noise that makes the rest of the game less compelling. Combine these two, and you have the basis of balance. Specific balance issues are listed below:
1) Dominant Strategies
A dominant strategy is a case where one choice is always better than the other. This is always bad, because its effectively one choice, with noise -- not a compelling bit of gameplay. In multiplayer games, this manifests itself as times where you always play the same strategy no matter what, or when one player can play a particular sequence of actions and responses that guarantee the other player will be at a disadvantage, no matter what.
2) The true "costs" of actions
Actions in most games have some inherent game-based cost, be it some resource, like gold or minerals, or in the case of many turn based strategy games, a turn, as well as some player time investment. There is also an inherent player time investment that is especially relevant in real time games, and somewhat (though not as) relevant in other games. Actions with low payoff, low game cost, but high player time investment, are generally less useful (and imbalanced-weak) compared to low payoff, mid game cost, low player time investment actions, simply because many players will refuse to "waste" a lot of time, or in the case of a real-time game, cannont afford to use such time in that way.
Several examples from Starcraft:
An example of this is mass zergling wave rushing being favored over a number of other more "cost effective" strategies. You could also consider the fact that a mass tank push vs protoss is theoretically unbeatable, but in practice, pretty easy to beat because of its difficulty of flawless implementation.
Its important to realize that actions have a concentration cost AND a game cost when dealing with real time games especially. You must balance both to keep all actions equally appealing in their own unique way.
3) Risks should correlate to Rewards, or Versatality is power
When an action is versatile, its allowing you to in effect make several actions, for the cost of one (or at least the concentration cost of one). It allows you to not use as much time when responding to a variety of things, and allows you respond to indeterminate situations with less risk... This can be a substantial benefit in many cases, and versatile actions that allow an adverse response from the other player (i.e. an attempted counter) should not be especially strong compared to less versatile actions, since versatility is "playing it safe".
4) Balance across skill levels
When dealing with a competitive game, you want skill differentiation -- when you have skill differentiation it is important to have balance be constant over skill levels for the most part. In this respect, it is important that as skill increases, that various game components increases in their apparent effectiveness. One potential hazard here is that in extreme situations, the game might just not be fun at all at a particular skill level.
5) Scenario Imbalances
Sometimes called "map imbalances", these are imbalances stemming from the initial conditions of the game. Sometimes, when you can start with different options individually, the "global" configuration of the game may be slanted towards favoring one particular individual configuration. Similarly, sometimes there may be only one global configuration, but specific starting "positions" or states might be better than others. As the alternative name implies, this is most apparent in situations where you have a map.
6) Dominant Playstyles
Similar to Dominant Strategies. Often, when playtesting your game, you will notice that there are several major playstyles which are at least somewhat feasible. Its important to try to cultivate as many playstyles as you realistically can (don't expect more than a few though), and to get them to the point where they are about equal in potency.
The distinction between dominant playstyles and strategies is that I define playstyles as being a more general approach to the game, and strategies as being specific responses to situations. In some games they are one and the same In some they are not.
7) Single-Player Games and Difficulty Balance
You don't want single player games to be too hard. In general, a good way to balance them is to balance them to slightly below the median skill level, and then add extra goodies for higher skill level people, or just add higher difficulty settings. It's also important to make sure that the game becomes progressively more difficult. Mech Commander is an example of a game with a big problem in this department, in that mission 3 out of 25, also known as the farm mission, was the hardest mission in the game, and prevented a lot of people from playing past the very beginning of the game. Don't do this!
8) Expected Return of Game Elements
You obviously should balance the power of things largely based on the cost to get them, and the value of the effect they can create. When dealing with something that becomes obsolete (which should be avoided, but has to happen in some types of games), consider the expected return over time, including the period after it grows obsolete. For instance, in a real-time strategy game, a unit that is good for the first 5 or 6 minutes, but that becomes obslete after, can be a bad thing, because normally when your units survive you come out ahead, but if you fight early on with this unit, and have survivors, its as good as dead as the game progresses...
9) The Basic Cost Effectiveness Equation
Cost effectiveness is Square Root of [Firepower*Hitpoints/Cost-squared]
Test it if you don't believe me. Firepower represents game effect, be it damage, or something else. Hitpoints represent the durability of the game element, which generally will be just that -- hitpoints. Cost is the cost it takes in the game, usually in terms of a game resource. Its harder to measure concentration of course.
Theres other equations too, but we wont cover them now, and I want to avoid too much vigorous hand-waving.
Implementing play balance
With all that theory aside, we can think about acutally balancing agmes. Play balance is one of the more difficult aspects of game design, and one that has broken all too many otherwise good games. Few designers consider the implications of their game in terms of play balance in initial design stages, and only a few more even think about it prior to a mid-alpha state of development, especially when dealing with computer games. The key, like many things in game design, is to think things through early so that you dont have a hellish task to complete later on. Making a system innately balanceable is the first step to making a balanced game -- some systems will be easier to balance than others, and some will be almost completely impossible to balance acceptably well. This lecture discusses how to make your games more balanceable, for the most part.
1) Modularize, Modularize, MODULARIZE!
Its important to modularize your design, as the title mentions. What I mean by this is breaking your game elements into seperate pieces, and trying to ensure that the play balance values of each influences only one game dynamic a piece. Theoretically, if you tweak the value, you then only influence but one thing's play balance state. An example of this going awry would be the Mutalisk in starcraft. It has two major purposes, those being anti-air primary attack and anti-land skirmishing. It shares a common attack for both purposes, as well as hitpoints, etc. Because of this, Blizzard had _extreme_ difficulty balancing it satisfactorily because to change the value of the attack or hitpoints influenced BOTH roles, rather than the problem role. (i.e. the land attack was consistantly too much, but if they downgraded it too much, they ended up with the unit being too weak against air... The Blizzardism for modularization (which is also good for hypnotizing impressionable business folks) is "Purity of Purpose", at least more or less.
2) Unnecessary Details are... Unnecessary
Dont add details that you dont need. Again, the purer the game design, the easier it will be to balance it. "Noise" in your design will probabaly undermine other factors too, and its hard to play balance when its hard to determine what variables are actually the significant ones. The more easily understood the game system is, the higher your chances of figuring it out well enough to balance it. When I was an inexperienced designer, I'd often add in details without thinking too much about it, and end up having to gut the system later on and trim off fat to get it play balanceable.
The Play Balance Process
The preferred approach to game balance, in my opinion, is:
1) Work out "balance on paper" and think through situations and counter situations. Try to get a feel for how you want things to work in general.
2) Balance the game for a wide but very select group of game elements, primarily BY FEEL. Exclude math at this point. Ensure that the game is balanceable as well, namely, that hooks to affect major game dynamics are in place.
3) Scale the balanceu from specific elements to similar elements using balance math, such as the fragmentation equation, and the effectiveness equations.
4) DO NOT rely on master math formulas, etc. You could write a PHD on this stuff, and theres always weird variables you wont account for. You CAN do a board game in pure math, but its still hard. Once you use math to scale things into relevant ballparks, go back again to your design instincts, and do a rigorous qualitative test to discover if things are working against other game elements as designed. This will tell you 100x more things than most balance math will.
Take Small Steps!
Inexperienced designers make radical changes to their play balance #s with every revision of the game. Don't be this designer! After your intial 1 or 2 play balance sweeps from your initial qualitatively balanced set of elements, take it slow! Although well-designed and modularized systems will resist this, most game systems tend to have balance #s that are easy to "overshoot". If you dont take baby steps, changing one or a very few elements in very small amounts at a time, it will be hard to seperate out the effects of doing these things. Also, when attacking a specific problem, try just tweaking ONE quantity at a time.
Players Play Consistantly
If you have an imbalance in one stage of your development, your playtesters will continue to play with a playstyle that reflects it. Give playtesters time to adapt to changes in your state of balance, and strongly encourage them to try different approach. Top notch QA people will do this without being asked, but most of them either get other jobs, or get promoted to design, so you won't have many top notch QA people. Similarly, fans are inexperienced testers and wont know this either. Be careful!
Dont Be Afraid to Force What You Want!
If you want something to be more effective, MAKE IT MORE EFFECTIVE. Sometimes you try and try to work around something to get a result by altering indirect contributors. Don't do this too much when a simple straightforward approach will work.
An example of this in my experience developing was when I wanted to be able to have certain units be countered strongly by certain other units, and yet strong against other units. I tried to make this work through movement rates, and ranges, and a lot of other things... I also had an armor system that had some semblance of reality, but was primarily a game design construct. After 8 or 9 hours of wasted effort, I just made a system that applies an effectiveness coefficient to every units attack, vs any other unit (i.e. its a table of values for every unit vs every unit), which gets me EXACTLY the effect I want, without screwing up tons of other things by altering indirect forces on this. Of course, those indirect forces still cause things to work a certain way (like high speed, moderate range units eat slow melee units), but I can close the gap with anything that needs such a change by changing a single number -- which is completely and utterly modularized.
The downside here of course is that doing too many "Secret modifiers" or whatever you want to call them tends to make the game less intuitive. I got around this by pressing my modifiers in directions that the units were "supposed" to work anyway.
Workshop:
1) Study how chess pieces are balanced relative to each other. Why are the strongest pieces the strongest, and how might you analytically measure chess piece effectiveness?
2) Is monopoly a balanced game? If not, suggest some fixes. Otherwise, suggest why it is in fact balanced.
~SDC~
Stardock Forums
)
). It IS a civ game after all, and real-life civilization IS kinda complicated.