The Tangled Concept of Balance

January 6, 2008 at 1:40 am (Game Balance, Game Design, Philosophy)

Game balance is one of those terms that is used frequently in discussion, but difficult to define.  Most people seem to have a general agreement about what the term means, but articulating it is problematic.  People agree that a fighting game where a particular character consistently beats another is unbalanced, based only on that one fact, but the same people will also agree that a game of Rock-Paper-Scissors is balanced even though paper has a 100% win rate against rock.  People agree there is a balance problem in a RTS game where a particular unit is too weak to justify its use in a competitive game, but see no problem in the fact that many combinations of units are too weak to ever be used effectively.  How can this be consistent?

Of course, there are also cases where people flatly disagree about what is and is not balanced.  Sometimes these are disagreements about the particulars of a game, but sometimes they seem to be disagreements about the theoretical meaning of “balanced.”  How are we to discuss game balance in such cases?

Is there even a common concept underlying all our ideas of balance?

I’ve wondered about this for a long time–years, in fact.  I’ve tried and discarded theory after theory.  At long last, I think I have a satisfactory answer.

A Technical Point

Of course, I could just rattle off any arbitrary definition and claim that’s what “game balance” is, but there’s not much point in that, is there?  We want a definition that means pretty much what everyone already means when they say “balance.”

But at the same time, we don’t want (and can’t have) a perfect agreement with what everyone means, because not everyone means exactly the same thing, and not everyone is really sure what they mean.  We also want “game balance” to be a useful concept–for it to mean something that is in some way identifiable, and relevant to game design–since, otherwise, why bother?

So let me admit up front that what I’m about to present as “game balance” is probably not exactly what you think “game balance” means right now, and it’s certainly not what everyone has ever meant by the term.  But I think it’s roughly in line with common usage, and I think it’s a clear and useful concept, once we get down to it.

Balance vs. Fairness

The first concept generally associated with game balance is what I’m going to call fairness: an equal chance of victory given to all parties.  A fair game is one in which no player begins at an inherent advantage or disadvantage, except due to intrinsic qualities of the particular players.  Thus, a more experienced or skilled player may be more likely to win than a novice player, but there is no inherent advantage or disadvantage to being assigned a particular player slot (for example, we would say chess is fair if there were no inherent advantage in playing white or black).

Fairness, as I’m definining it, allows that players may make choices which are more or less advantageous to themselves, and that the game’s outcome may be affected by random events that favor one player or another.  However, before the game starts, there is nothing about the game that will help you to predict the eventual winner.

Note that a game being symmetric with respect to all players is a sufficient condition for it to be fair (as all players will have exactly the same options and resources).  Symmetry is not a necessary condition for fairness, but perfect fairness is exceptionally difficult to achieve without symmetry.  We can allow for varying degrees of fairness, measured by a priori certainty that a given side will win, and say that games that more strongly favor a particular side are less fair.

Take a moment to think about what games you have played are or are not symmetric; you may be surprised.  Chess, for example, is not symmetric; only one player has the first move.  For the same reason, checkers, tic-tac-toe, bridge, uno, and monopoly are all asymmetric.  WarCraft or StarCraft, on the other hand is symmetric (if played on a symmetric map): all players have equal opportunity (at the start of the game) to choose between all starting races, all units, all technologies, etc. and they all act simultaneously with equal knowledge of each other.  Street Fighter and similar games are also (usually) symmetric, as long as you consider character selection a part of the game.  So this definition allows us to trivially prove that WarCraft is 100% perfectly fair (at least as an abstract game) and chess is probably not.

A few people will contend that fairness, as I have defined it, is exactly the same thing as game balance.  Most will say that it is a part of balance, or that it is one special case of balance, but that balance encompasses equivalence of certain player choices as well as player starting positions.  I am going to take a much more radical position and argue that fairness is not even a part of game balance.

Allow me to say that again, to be sure it sinks in:  I am arguing that a game can be balanced without being fair.

This probably strikes you as ridiculous, and that’s understandable.  Surely, you protest, if one took an example game like chess or StarCraft and discovered some new unfairness to it, I would consider the game to be less balanced as a result?  Indeed, I certainly would.  But not based solely on the fact that the game is unfair.

Let me propose a counter-example.  Suppose someone designs a game where one side is intended to win more often than the other–perhaps it’s a many-on-one game, and the designer wants the “many” side to win more often, on the theory that a game where most players lose most of their games will be less appealing than a game where most players usually win.  Suppose she carefully calibrates various different game mechanics and score thresholds and random factors and ends up with a game where one side wins 70% of the time and the other wins 30% of the time, with those percentages consistent across different skill levels, strategies, and play styles.  By the above definition, this game is unfair.  Are you sure you’re prepared to say that it’s unbalanced?

Let’s take a more common example.  Suppose we create a game that is otherwise fair and balanced, and then we add a “handicap” feature that allows one player to specifically start at an advantage or disadvantage.  If an experienced player plays against a novice, and they adjust the handicaps somehow so that each of them wins half the time, are you not inclined to say that their game is balanced?  If the same handicap settings are used for equally skilled players, causing one to win much more often than the other, are you not inclined to say that the match-up is now unbalanced?  More importantly, to whom would you ascribe fault for unbalancing the game–the designer, for adding the handicap feature, or the players, for miscalibrating it?  I think most people would intuitively want to say that the designer produced a balanced game and that the players have unbalanced it by adjusting the settings:  I think this is an important clue to what we really mean by “balance.”

Parity of Initial Options

I will return to my previous point, but let’s consider some other views of game balance.  One might be inclined to view game balance as having to do largely with stylistic options players select at the start of a game (or arguably, before the game begins).  After all, most of the complaints regarding game balance seem to be about such:  the different races or factions of a strategy game, the different characters of a fighting game, the different classes in a role-playing game, and so forth.  All of these seem like choices which change the style in which the game is played, and which the player should be able to choose freely without concern for hobbling himself; we expect that they be “balanced” against each other, so that any of them can win against any of the others.

There are, of course, cases like rock-paper-scissors, where your initial choice is the only choice you make, and some options always defeat others.  We could allow certain pairings to be slanted as long as the overall success rate of each choice is equal, but this suggests (for example) that we would call StarCraft balanced even if the Terrans always defeated the Zerg, the Zerg always defeated the Protoss, and the Protoss always defeated the Terrans; that seems rather unlikely.  Instead, we would probably tend to say that it’s OK in rock-paper-scissors because your first choice is your only choice; it is unacceptable in StarCraft, Street Fighter, or WoW because it is the first choice of many in a long game.  Besides, we probably want to expand it to include the first two choices, or the first several choices, in some games–perhaps the issue is that we don’t want the game to be decided too early?

While I agree that we generally don’t want the game to be decided too early, the more pressing problem is that this definition fails to encompass many textbook “imbalances.”  We want to call StarCraft imbalanced if there is never a good reason for Terrans to build firebats, even if Terrans have the same win rate as other factions.  We would call Street Fighter imbalanced if it was stupid to perform any attack other than a throw.  These choices occur throughout the game, not just at the beginning, and these issues do not give any particular player an advantage over another.  Focusing only on the start of the game is taking us in the wrong direction.

Strategic Equivalence

Another proposal would suggest that what we want is for the player’s chances of winning not to be influenced by his choice of strategy; all available strategies should be equally effective.  Thus, we are justified in complaining that a particular unit or attack is too strong or too weak, because strategies focusing on or ignoring that particular option will be more or less effective, thus disrupting the game’s balance.  Simultaneously, we can still explain why we want all factions, characters, or what-have-you to possess equal chances of victory.

But it’s quite hard to defend this definition if someone asks the simple question, “any strategy?”  Because, no, off course we don’t mean that you’ve got an equal chance to win with any strategy.  A player who stands there and does nothing can be expected to lose, and this does not invalidate the game’s balance.  That a player can lose a game of chess by refusing ever to capture an enemy peace does not prove the game imbalanced.  We’re expecting some sense here.

In fact, we don’t want complete equivalence even between the “good” strategies.  We want to be able to say that one player played better than another because he used a better strategy, and that’s just not possible if all strategies are equally good.  Playing a game whose outcome isn’t influenced by your decisions is boring; we want our strategic choices to have a very significant and direct impact on the game.

If we think about it generally, this theory of game balance sounds about right.  It allows us easily to recognize most complaints regarding game balance.  But as soon as we try to apply it rigorously, it collapses like a house of cards.  We can point out counter-examples left and right; in fact, I don’t think I can find any positive examples, where a game we like actually meets this criteria.  There’s still some crucial element we’re missing.

Stylistic Agnosticism

Coming at the problem of game balance from another angle, we might say that the problem isn’t that some choices change our chances of winning, but that the wrong choices change our chances of winning.

Many games have choices which are clearly intended to be stylistic, rather than strategic.  The decision of which faction or character you choose at the start of a game is supposed to be based on your personal perferences or playing style, not on how likely each is to win.  The choice of focusing on swords or spells, on military might or economic growth, these all–depending on the game–are supposed to generate different play experiences, not to be the choice that wins or loses the game, and thus all of them should be valid.  In fact, we should have a clear separation between stylistic choices and strategic choices, and refrain from inflicting strategic penalties on stylistic choices, or stylistic constraints on strategic choices.  That way, we expose players to the maximum breadth of game experiences without bestowing any unfair advantage.

Of course, many games do, in fact, mix style and strategy.  Choosing to play as one faction opens some strategic options and denies you others.  Perhaps this is an evil that should be minimized?

I don’t think so; that would imply that our stylistic choices are purely aesthetic.  In order for them to change how the game is played, rather than just how the game looks, they need to affect your strategic options in some way.

Besides, this implies that a “balanced” game can have any strategic properties you care to name, just as long as they aren’t intertangled with stylistic options.  I can have a game with two attacks that are precisely identical, except that one of them always improves your chances of winning the game more than the other, and you’d have to call it “balanced,” because there’s no stylistic issue involved.  This cannot possibly be all that people mean by game balance.

Equilibrium and Intent

So here’s what we’ve covered:

  • We seem to expect all factions, characters, classes, colors, etc. to present equal opportunity of victory.
  • We seem to expect all units, attacks, spells, items, etc. to be useful under some circumstances.
  • We think that game balance requires some sort of vague parity between different strategies, but not that all strategies are equally good.
  • In fact, we think that some strategies can be obviously and hopelessly bad, and this doesn’t even make us worry about the game’s balance.
  • We think that the use of a handicap doesn’t necessarily mean a game is unbalanced.
  • We think that the misapplication of a handicap means that the players have unbalanced the game, not that the designer has committed some error.

In order to find commonality in these points, we’re going to need to step back from the particular cases we’ve analyzed and look at the game from a more abstract perspective.  What is it about these particular game mechanics that gives rise to these expectations?

And so, at last, we come to my proposed definition:

game balance (n.) – The degree to which the rules of a game cause players who are attempting to win the game to play in the manner intended by the game’s designer; the degree to which effective gameplay agrees with intended gameplay.

…at this point, I imagine several of my readers will be offended.  Having spent paragraph upon paragraph disparaging traditional understandings of game balance, this is all I have to present?  No reference to anyone’s chances of winning?  No relationship between different strategies?  Not the slighest hint of style?  And I have the temerity to define it relative to intent, allowing any game to be “balanced” just because the designer claims that’s how it’s supposed to work?  How dare I!  How could this be what anyone in the world but me means by “game balance?”  What use could this possibly be in any informed discussion?

But it’s not so bad as you think, I assure you.  Technically, this definition allows anything to be balanced if it was intended, but if someone produces a game we would normally call unbalanced and claims it works as intended, we can still respond “in that case, your design sucks.”

And while we don’t necessarily know exactly what was intended in all cases, most important design goals are easily inferable from the structure and presentation of a game.  We expect that every unit or ability is intended to be used under some circumstance, because the designer has decided it is worth the trouble of designing, implementing, and presenting to the player, and something that isn’t intended to affect the game wouldn’t be worth that trouble.  We expect the choice of faction or character to be mostly stylistic because that’s an accepted convention (and widely considered a good idea).  A game which devotes a significant portion of the user interface to stealth indicators clearly intends for stealth to be a major component of the gameplay; a game with a health bar intends that the player not generally die from a single hit; and so forth.

This definition also cleanly resolves a lot of thorny distinctions in what we consider imbalanced.  We consider a single unit or ability that is never useful to be a balance problem, but a certain combination of units or abilities that isn’t ever effective doesn’t cause us to leap to the same conclusion, because we have no indication that that combination is intended to be used (once all the pieces are in the game, it actually would have taken more effort to forbid the combination than to allow it).  We consider most games to be poorly balanced if we discover they are unfair, because fair games are the normal convention and are usually more interesting, but when a game includes a handicap system or some other element that clearly indicates unfairness in design intent, we don’t consider it a flaw.  Additionally, we consider a particular match-up balanced when the players set handicaps to give all players an equal chance of winning, and consider it unbalanced when the handicaps are not so calibrated–but consider the imbalance to be the players’ fault–because we believe the players are intending to calibrate the game in order to give all parties an equal chance.

Even more, this definition gives us the power to explain seeming imbalances that don’t fall into any of the usual categories.  For example, suppose there’s a strategy game where players can build walls, and playtesting reveals that defending your base with walls is ineffective, but walls are useful for some bizarre tactic, like tricking the AI into sending its armies to remote, unimportant locations where you build them.  We can easily imagine that the game is fair, that major strategies are roughly equally effective, and we see that walls still have a solid tactical function that allows them to be useful.  However, I still want to say that there’s a balance problem–and I suspect you do, too–because walls are not effective at what they’re supposed to do, and they’re too effective at something they’re not supposed to be good for (their intended functionality being fairly easily inferrable from the fact that they’re called “walls,” and perhaps from the game’s documentation).  There is no objective reason that having walls in the game is somehow better than having decoys; there is not necessarily any claim that someone has an unfair advantage, that the strategic depth of the game has been reduced, or anything else of the sort.  The only problem here is that, in order to play well, players must play the game differently than intended.

This concept puts the “balance” metaphor in a new light.  Conventionally, we have thought of game balance as weighing different players or strategies, as if on a scale, to determine which is better.  This definition invites us to think of the game itself as being some standing structure, and balance as its resistance to tipping or collapsing when we subject it to the stresses of players trying to win.  When we poke and jostle the game, does it break, or fall over into some degenerate state, where the gameplay is no longer interesting–or does it wobble and then right itself, returning to the original set of strategies and trade-offs than the designer built in?

Using this definition, we may occasionally run into situations where there is a legitimate disagreement about whether something is a balance problem or just plain bad design, but I’m not so worried about that.  From a criticism standpoint, it doesn’t matter which it is; from a designer’s standpoint, you should easily tell the difference.

Some might fear that, while the definition does appear to encompass all the key things we want to call “balance problems,” that it is too broad–that we’re measuring the overall quality of the game, instead of one specific category (or that we will fall into doing so after we resolve the bad balance/bad design issue).  But I don’t think this is true:  while game balance touches on pretty much all parts of the gameplay, I think we can still meaningfully discuss a game’s depth, customizability, replayability, user interface, and fairness as considerations that extend well beyond game balance.  Additionally, things like the game’s art, story, and style are clearly separate from balance considerations.

There are, of course, a few cases where this changes our conventional understanding of what is balanced or imbalanced.  Most people have traditionally called all unfair games unbalanced.  Many people would tend to say that games where a single strategy is usually clearly the best “unbalanced,” without further consideration.  I’ve seen many different definitions of game balance proposed in different books, web pages, and other forums, and not a single one (that I have seen) agrees with mine.  Some designers speak of deliberately designing imbalances into a game, which clearly conflicts with my definition.

I have attempted to argue that these rules-of-thumb are usually true, and can be derived (for most cases) from my definition, but that exceptions must be allowed.  I have attempted to argue that these other definitions do not agree with our intuitive understandings of balance in all cases.  And I think that talk of deliberately designing imbalance arises as a form of opposition to these faulty definitions, rather than as an indication that we think game balance has somehow gotten too good.

Perhaps you disagree with me.  Quite likely, many of you find yourselves uncomfortable with this definition, if only due to its strangeness.  But I think it does a remarkably good job of explaining lots of seemingly weird cases that cannot be elegantly explained under any of the other definitions, and it seems to agree quite closely with most of the implied assumptions I hear (and make) when discussing game balance.

I think the idea of balance as stability–the ability of the game to hold its intended shape under duress–is very powerful.  It has the potency of a simple but far-reaching idea; very quickly, one can begin to see traditional balance problems as being things that derail or deform the game, rather than as being disruptions of the chances of a player winning.  The probability of victory (or more specifically, the player’s desire to increase it) becomes the force we are defending against, rather than the goal–and suddenly the exceptions to all the other definitions seem to fit into the grand pattern.

Though I’ve been relying on other definitions of balance for quite a long time–I like this one.  With a little reflection, I suspect you will, too.



  1. Jordan (JJ10DMAN) said,

    Excellent essay! The frankness , analogies, and study of specific examples makes for a clear and engaging read.

    However I think that using the intent of the game’s designer as a factor in the definition artificially brings in deeper the problem of differentiating bad design from bad balance, as well as making it far less useful as a definition.

    Instead I would focus more on the inference itself; it is not the designer’s intent we should use as a metric, but what the average player infers to be the designer’s intent. After all, unless the designer is somehow directly having a conversation with the player, that’s all that really matters to the player.

    In this way we can understand bad balance and bad design (with regards to balance) as two sides of the same coin; balance is the meeting of expectations, and design is the establishment of expectations. In an RTS, units are expected to have some purpose. If one unit is identical in every way to another, but has less health (and we assume there’s no strange abilities where less health can be a good thing), then that unit has no purpose; this is bad balance because it is a violation of the inferred expectations. What the designer intended literally has nothing to do with it.

  2. spellman23 said,

    Hm… Love the article, but I’m feeling a little down now.

    You have given a very nice presentation on the fallacies of people attempting certain types of balance, and given a nice little definition of balance, but now the question becomes, NOW WHAT?

    Sure we’re leaving plenty of room for any body’s idea of balancing their game the way they want, but in some respects it doesn’t answer the big question: how do we balance things? There’s a reason patches come out for games to fix balance all the time, tweak settings for competitive games, and so on. Sure, maybe it’s the designer’s own misjudgment of the players, and yes many times the patch is there to fix an exploit, but where do we go from here? Must the competitive gamers work within the designed frame of the game designer, or in reality are not the game designers also balancing to the players?

  3. Tim C said,

    This guy has some interesting thoughts on practical applications of fairness. Thought you might find it interesting.

  4. Sirlin said,

    I completely disagree with your definition of balance. It’s not useful at all. Designer intent is totally irrelevant to consider.

    Just to name one example, the fighting game Marvel vs. Capcom 2 has held up to tournament play for almost a decade now. It has a 54 characters, about 10 of which are solid enough to appear frequently in tournaments. Now, you might say that’s too many bad characters (yeah it is), but having 10 wildly different characters all hold up as viable after a decade of tournaments is pretty amazing. It’s ludicrous to say that game is not balanced because it’s not played the way the designers intend. I can virtually guarantee that it’s played vastly differently than anyone ever intended. But so what?

    Designer intent doesn’t matter. Only the game players have, as it stands, matters.

    Also, friendly advice (really, meant constructively), consider cutting the length of a post like this by 1/3rd, while containing the same number of ideas.

    (Lead Designer, Street Fighter HD Remix)

  5. Antistone said,

    Even if I completely agree about the impressiveness of your example, Sirlin, I don’t see how that breaks my definition, or why you say my definition is useless. I have not argued that balance is easy, or that you can’t have different degrees of balance, or that balance is the only important thing in a game.

    From your writings, Sirlin, you’ve always struck me as someone who is deeply concerned with how to win in tournament-level play and much less concerned with actual game design. I would agree that, from the perspective of “playing to win,” the designer’s intent is totally irrelevant–you’re trying to win the game in front of you, not the one in his head. And that means that you care about what I’ve called “fariness,” but not at all about wasted potential, unused portions of the game, or degenerate strategies. In fact, while you CARE whether the game is fair, fairness isn’t even necessarily a goal–you would presumably prefer the game to be stacked in your favor.

    But that’s not a typical perspective. Designers, critics, and even most players spend a considerable amount of time concerned with how WELL the game works, not just HOW it works. There’s a point other than who wins and who loses–otherwise we’d just flip a coin. And if you’re evaluating how well-designed a game is, the fact that 80% of the content is wasted is a pretty darn important consideration (though in your example, I suspect there wouldn’t be 10 viable characters if the game only included 10 in the first place–the designers probably threw in a ton of content because they knew they couldn’t correctly guess which characters would turn out to be viable in tournaments 10 years later).

    Allow me to offer an analogy. In software development, there’s a concept called a “bug,” which means that a program reacts incorrectly to a particular input. See, there’s a specification for what the program is supposed to do (separate from the program itself), either express or implied, and the programmer has to take that specification and craft his code to match it, filling in any details that were omitted from the original spec, but remaining consistent with all of the design goals.

    If you’re only concerned with using the software, it doesn’t matter to you what’s a bug and what’s not–all you care about is what the software actually does, and how you can use that functionality effectively. But if you are a designer, programmer, tester, or any kind of critic, “bug” is an extremely important concept.

    When a game is created, someone comes up with an idea for the game, and what it’s like to play it. He or she creates a specification for the game (express or implied) saying, in general terms, what it’s like. Then, someone else (or possibly the same person) needs to convert that idea into a set of concrete game rules that spell out exactly how the game reacts to each set of circumstances. This person has to fill in a lot of details that weren’t explicit in the original specification, while still maintaining all of the design goals.

    Just as in software creation, it’s not as simple as that–there’s back-and-forth, revisions and clarifications to the spec. Sometimes goals change, new concerns overriding old ones, and so forth. But it’s still extremely useful to distinguish between the original gameplay concept and the fully realized design, just as it’s useful to distinguish between a program’s specifications and its implementation. A discrepancy between specification and implementation is a bug. A discrepancy between a game’s vision and how competitive players actually play it is an imbalance.

    If you’re playing a video game at a tournament level, you likely don’t care which of the game’s responses are bugs and which are not. Unless some game function has been explicitly banned, if winning is your only goal, you’re just going to consider how you could use it (or how it could be used against you), not whether it’s “correct.” In the same way, you don’t care about whether some feature is balanced, you only care how you can use it.

    But that’s a special case. The concepts of bugs and imbalances are still extremely useful to a lot of people. In fact, one of the most important applications of those ideas is if you’re trying to design the game that’s going to stand up to the tournament players–they may not personally care about these concepts, but if the designers of their games didn’t, they wouldn’t have any good games to play.

    Also, friendly advice (really, meant constructively), consider cutting the length of some of your articles on by 1/3rd, while containing the same number of ideas. When I read them, I tend to get the feeling that you’re using an excessive number of examples to make an obvious point.

    On the other hand, I also end up disagreeing with a sizable fraction of your conclusions, so maybe you’re using the right number of words, and just spending them on the wrong arguments.

  6. Tommi said,

    game balance (n.) – The degree to which the rules of a game cause players who are attempting to win the game to play in the manner intended by the game’s designer; the degree to which effective gameplay agrees with intended gameplay

    Personally, I’d call that the definition of good design: A system that provides the output it is supposed to provide.

    That said, semantic arguments about what is the “right” way to define a word are usually unproductive; it is much more interesting to define a word in some particular way and what the implications are. YMMV.

    In the world of roleplaying games, which are generally not competitive, game balance roughly means the lack of bunk choices and that characters work in the way they should work; character that is tough should be hard to take out with physical damage, for example.

  7. Scanning the blogosphere « T.A Community said,

    […] By fing0lfin Categories: Uncategorized Tags: article, balance, rts, strategy “The Tangled Concept of Balance” – A superb article over at Gaming’s Alembic about balance, fairness, strategy and style with […]

  8. fing0lfin said,

    Really great post. Check out my (very rough!) early notes about a futuristic RTS game concept I’m working on over – perhaps you can check it out if you have time. I’d be interested to hear your opinion on the feasibility of such a game (especially in terms of balance/mechanics – I’m no programmer for sure!).
    I also put a link to this article in my blog, hope you don’t mind?!

  9. Scanning the blogosphere « The Future of RTS said,

    […] the blogosphere “The Tangled Concept of Balance” – A superb article over at Gaming’s Alembic about balance, fairness, strategy and style with […]

  10. Eldar said,

    Good grief, “Sirlin” is a moron. Excellent rebuttal to him, Mr. Antistone, and I find it hilarious that he never responded.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: