Level 16: Game Balance

When veteran gamers or game designers are playing a game, if they are doing too well or too poorly, they will often comment on the game’s balance. This word is important, but I fear it is often overused. Like the word “fun,” there are different kinds of balance, and understanding what game balance is and why it’s important is what we cover today.

Why are we only covering this now and not earlier (like, say, at the start of the Design Project)? As mentioned earlier, balancing the game is something that is best left until after you have a good set of core mechanics. Balancing a game that is simply not meeting its design goals is a waste of time, and when you change the core mechanics you’ll just have to balance the game again. So here we are, with a work-in-progress that has survived multiple rounds of playtesting, and it is time to take it to the next level.

About the Pace of this Course

At this point, I realize from the comments and some forum posts that many of you are falling behind. As a reminder, this course is going at a brisk pace, and I recognize that many participants may not have the time to devote to this one hundred percent. On the bright side, the playtesting part of designing a game can go at a fast or slow pace; for your own hobby projects, time may not be of the essence, so you can continue to follow along at your own pace.

If you can still keep on schedule, that is great. If you are finding that setting up playtest sessions and modifying your game is taking too much time, my suggestion would be to do things as soon as you are able, and then follow along through the rest of this course a little behind schedule. I’m not taking this content down any time soon, so it will patiently wait for you. The rest of the course will concentrate on the Design Project and there will be no other homeplays to distract you.

I would recommend keeping up with the readings on this blog, however, if you have the time. This will give you an idea of what to look for in your project as you look ahead, and I may make announcements here that I’ll want you to see.

Readings

No additional readings for today. There is plenty of material here in this blog post.

What is Game Balance?

In a two-player game, saying it is “balanced” usually means that one player does not have an unfair advantage over the other. But we also hear the term used in relation to single-player games, where there is no opponent other than the game itself, and any “unfair advantage” the game may have could just be considered a challenge. And then we may talk of individual cards in a game like Magic: the Gathering as being “balanced” even when all players have access to that card, so it does not give an advantage to any individual. What’s going on here?

In my experience, when we talk of “game balance” we are generally talking about one of four things. Context usually makes it clear which one we are talking about:

  1. In single-player games, we use “balance” to describe whether the challenge level is appropriate to the audience;
  2. In multi-player games where there is asymmetry (that is, where players do not start with exactly equal positions and resources), we use “balance” to describe whether one starting position is easier to win with than another.
  3. Within a game, if there are multiple strategies or paths to victory that can be followed within the game, we use “balance” to describe whether following one strategy is better or worse than following another.
  4. Within a system that has several similar game objects (such as cards in a trading-card game, weapons in a role-playing game, and so on), we use “balance” to describe the objects themselves, specifically whether different objects have the same cost/benefit ratio.

Let us examine each of these more closely, and then we will go over some practical techniques for balancing your game.

Balance in Single-Player Games

In single-player games, we use “balance” to describe whether the challenge level is appropriate to the audience.

Note that, by simply playing the game and getting experience with it, your audience will eventually become more skilled at the game. This is one reason why the later levels of video games are usually harder than the earlier levels. (Recall that another reason is so that the gameplay matches the dramatic tension in the narrative.) The change in difficulty over time in a single game has a name: we call it pacing.

There is one obvious problem here that we face as game designers: how do we know what an “appropriate” challenge level is? Sure, we can say that a logic/puzzle game for adults is probably going to be harder than a similar game for young children, but beyond that… how are we supposed to know what is too easy or too hard? The obvious answer: playtest!

There is another problem, however: not all players are exactly the same. Even within a narrow target audience, players fall along a bell curve, with a few that will be highly skilled and a few that are the opposite. In your playtests, how do you know where your testers fall on that bell curve? If you are just starting out, the ideal thing to do is to use lots and lots of playtesters. When you have dozens or hundreds of people giving feedback on your game, you can get a pretty good idea of what the overall ranges are. As you gain experience as a game designer, you will get a better feel for your audience, and you will need progressively fewer and fewer playtesters to give you the same good results. (If you’re starting out but you don’t have the time or resources to do lots of playtests, you can sometimes fake it if you have some idea of where your own playtesters fall on the curve. Do they consider themselves above-average or below-average skill level, compared to the other kinds of people you’re making your game for?)

Even if you have a pretty good idea of how to modify the difficulty of your game and where your target audience falls, what do you choose as your challenge level when the audience is diverse? No matter what you do, your game will be too hard for some people and too easy for others, so this appears to be a no-win situation. If you must choose a single level of challenge, a rule of thumb is to aim for the middle of the curve, as you will get the most people (the widest possible audience) that way. Another way around this is to provide support for those at the ends of the curve, using multiple difficulty levels, handicaps, or alternate rule sets to make things easier or harder.

Balance in Asymmetric Games

In multi-player games where there is asymmetry (that is, where players do not start with exactly equal positions and resources), we use “balance” to describe whether one starting position is easier to win with than another.

Truly symmetric games are rare. We think of classic games like Chess and Go as symmetric, since each player starts with the same set of pieces and plays by the same rules, but there is one asymmetry: one player goes first! If you modify Chess so that both players secretly write a move on a piece of paper and then the moves are performed simultaneously, it becomes fully symmetric (and plays very differently).

This brings up an interesting question: if your game is symmetric, do you need to worry about game balance at all? After all, both players start with exactly identical resources and starting positions and so on, so by definition no player can have an unfair advantage. This is true, but the designer must still consider other types of balance, particularly whether there is a dominant strategy. Simply making all players equal does not get you off the hook.

Even if your game is asymmetric, why bother balancing it? The simple answer is that it is a typical player expectation that a game will not give an automatic advantage or disadvantage to certain players, other things being equal. (You can play around with this. The card game The Great Dalmutti, for example, intentionally casts players in unequal roles as a way of showing how life isn’t fair; but that is part of the game, and the instructions and mechanics go to great lengths to set player expectations accordingly. But if your game is not breaking this rule with specific intent, you should be thinking about how to make it as balanced as possible.)

Asymmetric games are, naturally, harder to balance. The more asymmetric, the more carefully the game must be playtested carefully. One of the easiest ways to confirm this kind of balance is to find ways of relating each players’ resources to one another. If you determine that in gameplay, one Apple is always worth exactly two Oranges, then a player who starts the game with an Apple will be balanced against a player starting with two Oranges.

Sometimes, players are so different that direct comparisons are impossible. Some games not only give players different starting resources and positions, but also different rules to play by. Some players may have exclusive access to certain resources or abilities. One common asymmetry in games is to give players different and conflicting objectives (for example, one team’s objective is to survive for some number of turns and the other team’s objective is to eliminate the first team before that many turns). The more difficult it is to make direct comparisons between players, the more you will have to playtest to compensate.

Balance between Strategies in a Game

Within a game, if there are multiple strategies or paths to victory that can be followed within the game, we use “balance” to describe whether following one strategy is better or worse than following another.

One might wonder, why bother with this? If a game allows for multiple strategies but one is more powerful than the others, doesn’t exploiting the best strategy just equate to players trying to win? As long as no individual player has an unfair advantage, isn’t it okay for your game to simply be “whoever finds the most powerful strategy wins”?

The problem here is that, once a dominant strategy is discovered, astute players will ignore all suboptimal strategies. Everything in the game that is not part of the dominant strategy becomes extraneous noise. There is nothing inherently wrong with a game that has a single winning strategy, but in this case the suboptimal ones should be removed to make the game more streamlined. If you include options for players that are suboptimal, these become false decisions, because really there is only one decision (follow the dominant strategy).

If it is worth including several potential winning strategies in a game, then, it becomes much more interesting if those strategies are balanced. Again, much of this comes down to playtesting. In this case, when players are playing your game, make note of whether certain strategies seem to be used more often than others, and which ones seem to win. If several items are available for players to purchase in a game, is there one that seems to always get bought early, while others seem to be used rarely if ever? If players have a choice of actions each turn, does the winner of each playtest always seem to be the one that chose one particular action more often than everyone else?

Playtesting alone is not automatic proof that a particular strategy is unbalanced, but it should give you strong signals that certain aspects of the game need closer inspection. Sometimes, players will use a particular strategy because it is the most obvious or the easiest and not because it is the most optimal. Some players will avoid anything that seems too complicated or requires finesse, even if it is ultimately better in the long run.

Balance Between Game Objects

Within a system that has several similar game objects (such as cards in a trading-card game, weapons in a role-playing game, and so on), we use “balance” to describe the objects themselves, specifically whether different objects have the same cost/benefit ratio.

This kind of balance is specific to games that give players a choice between different game objects. Some examples:

  • Cards in a trading-card game. Players build a deck with a set number of cards from their collections. The choice of which cards to add is one of the key factors in the game’s outcome, and designers try to make the cards balanced with one another.
  • Units in some war games and real-time strategy games. Players have the ability to purchase units during play, and different kinds of units may have different abilities, movement rates and combat strengths. The designer will try to make the units balanced with one another.
  • Weapons, items, magic spells, etc. in a role-playing game, either tabletop or computer/console. Players may purchase any of these for use in combat, and they have different costs and different stats and abilities. The designer will try to make these objects balanced with each other.

In all of these cases, there are two goals. The first is to prevent any game object from being so weak that it is useless in comparison with other objects. This again becomes a false choice for the player, because they might be able to gain or purchase a certain object but they will quickly find that it is not worth using; the object is therefore a waste of the player’s (and designer’s) time.

The second goal is to prevent a game object from being too powerful. Any single game object that becomes a dominant strategy makes all other objects in the game useless in comparison. In general, if you absolutely must choose between making an object too weak or too powerful, err on the side of making it too weak.

Two objects are balanced if they have the same cost/benefit ratio. That is, what you give up to gain access to an object (this includes explicit costs like in-game money or resources, and also opportunity costs like drawbacks, limitations, or exceptions to the object’s capabilities) should be in some proportion to the in-game benefits you get from the object. The costs and benefits do not have to be exactly the same (in fact, usually the benefits are greater, or else you would simply ignore the object). However, when comparing two different objects, the proportion of costs to benefits should be roughly the same for each.

Three Ways to Balance Game Objects: Transitive, Intransitive, and Fruity

I have encountered three general methods for balancing game objects. The first is technically referred to as a transitive relationship. In more colloquial (but still geeky) circles, it is called a cost curve. This is the most direct way to balance objects. The general idea is to find some desired proportion of costs to benefits. This may be a linear proportion (something twice as costly is exactly twice as powerful) or it may be curved in some way (perhaps there is a law of diminishing returns, where you have to pay progressively more for each additional bit of benefit; or perhaps there are increasing returns, where you essentially get a “bulk discount” for paying a lot at once). It all depends on your particular game, but playtesting, experimentation and instinct will help you to figure out what kind of relationship there should be.

The next step is to reduce every cost and every benefit to a single number that can be compared. Take all the costs of an object and add them together; also sum the benefits. Compare the two, and see if the object is giving the correct numerical benefit for the cost.

This method is often used in trading-card games. If the game has an established cost curve, it makes it much easier to create new cards with combinations of existing effects. In Magic: the Gathering, if you want to create a new creature with a given color, power, toughness, and set of standard abilities (say, a White 4/3 with Flying and First Strike), there are already several costs it can be, and designers working on this game (and sufficiently informed players) could tell you exactly what those equivalent costs are. Adding more abilities comes with an increase in cost, and decreasing the cost would necessitate a removal of stats or abilities.

The second method is an intransitive relationship between game objects, better known as a rock-paper-scissors relationship. In this case, there may not be a direct relationship between costs and benefits, but there is a relationship between the game objects themselves: some objects are inherently superior to others and inferior to still others. The game Rock-Paper-Scissors is the canonical example; none of the three throws is dominant, because each throw will draw with itself, beat one of the other throws and lose to the third one.

This can be seen in some strategy games as well. In many real-time strategy games, there is some kind of intransitive relationship between units. For example, one common relationship is that infantry are strong against archers, archers are strong against flying units, and fliers are strong against infantry. Part of the game is managing your particular allocation and positioning of units (in real time) in comparison to your opponent.

Note that transitive and intransitive relationships can be combined, as in the previous example. In typical real-time strategy games, units also have different costs, so a weak (but cheap) archer may still be defeated by a strong (and expensive) flying unit. Within a single class of units, there may be transitive relationships, but the different classes have intransitive relationships with one another.

Intransitive relationships can actually be solved, using matrices and some basic linear algebra. For example, the solution to rock-paper-scissors is that you expect the proportion of each throw to be equal to the others: there should be a 1:1:1 ratio. Now, suppose you modify the game slightly, so that each win with Rock scores 3 points, a win with Paper scores 2 points, and a win with Scissors scores 1 point. What is the expected ratio now? (It turns out the ratio is not what you’d expect; with optimal play on both sides, you would see 1 Rock for every 3 Paper and every 2 Scissors. The math required to do this is outside the scope of this course.) If you are looking for players to use objects in a certain proportion (with some being more commonly used than others), a well-balanced intransitive relationship is a good way to guarantee this.

A third method of balancing game objects is to make each one so different and unique from the others, that direct comparisons are impossible. (I call this “fruity” in the sense that the designer, and later the players, can only compare apples to oranges.) Since formal and numerical comparisons between objects cannot work, the only way to balance this is through excessive playtesting.

There are challenges associated with all three of these methods. For transitive relationships, everything relies on the designer finding the correct cost curve. If your math is wrong, it will be wrong for every object in the game; if you find one thing that is unbalanced, you’ll probably have to change everything. Transitive relationships are much easier to develop in retrospect after playtesting, than developing them ahead of time. Since so much relies on getting the math right, it also tends to take a lot of trial-and-error and therefore a lot of time.

Intransitive relationships, as noted above, take some tricky math to solve. Another drawback is that, unless done very carefully, their presence can make the entire game feel like glorified Rock-Paper-Scissors, which some players find to be a turnoff – many have the perception that intransitive relationships are nothing but guessing games, where every decision is based not on strategy but on luck and randomness. (A full discussion of whether this is or is not the case is also outside the scope of this course.)

“Fruity” relationships are really hard to balance, because one of the most important tools in doing so – mathematics – is no longer available.

Three General Game Balance Techniques

In general, there are three ways to balance games:

  • Use math. Create transitive or intransitive relationships in your game, and make sure that everything is in line with the cost.
  • Use your instincts as a game designer. Change the balance in the game until it “feels right” to you.
  • Use playtesting. Adjust the game based on the results of playtests, where the players are experienced gamers who have been instructed to play to exploit and win.

There are challenges with each of these ways:

  • Math is hard, and it can be incorrect. If your formulas are wrong, everything in the game may be off a little bit, which is inconvenient for rapid prototypes. Some really strange abilities or game objects may not have any math to them if they are too unique, requiring other ways of balancing.
  • Instinct is vulnerable to human error. It is also not absolute or reproducible; different designers may disagree on what is best for the game. This is particularly dangerous on large team projects, where one designer may leave in mid-project and another cannot take over (or rather, they can, but they will not be able to finish the game in the same way that the original designer would have).
  • Playtesting relies on the quality of your testers. Testers may not find every balance issue with the game; some problems will go undiscovered for months or years (even after public release of the game). Worse, some testers may intentionally avoid showing you rules exploits, because they plan to use them after the game is released!

What is a designer to do? Do the best you can, and understand both the strengths and limitations of the balance techniques that you are using. And as a game player, the next time you run into a game that seems horribly unbalanced, have some appreciation for how difficult it can be to get things perfect.

More Game Balance Techniques

Here are a few other random bits of advice I’ve picked up, in no particular order.

Be aware of the different objects and systems in your game and their relationships. You should already have done this during your initial design of the game, of course, but it is easy to forget the big picture when you start focusing on small details. There are two things in particular that you should return to first, whenever you make changes to your game:

  1. What is the core aesthetic of your game? Does this change support the core?
  2. Look at the interconnections between systems. If you change one thing, you should know what other things will be affected. Individual game elements rarely exist in a vacuum, and changing one thing can have ripple effects throughout the game. By being aware of the relationships between systems and objects, it becomes easier to predict the second-order effects of a mechanics change.

Make one change at a time. We’ve said this before, but it bears repeating. If something breaks after making a change, you know exactly why. If something breaks after making ten changes, you don’t know which change (or combination of changes) caused it.

Learn to love Excel. It can be any computer spreadsheet program, though Microsoft Excel is the most popular among game designers. Often, students look at me like I’m crazy when I suggest that a spreadsheet is useful in game design. (Like, aren’t those things only used by corporate finance dudes or something?) Here are some examples of how spreadsheets are used:

  • Excel makes it easy to keep lists of things and organize them. List all of your game objects and their stats. In a role-playing game, list all weapons, items and monsters; in a tabletop war game, list all units and their stats. Anything that you’d find in a reference chart in the instructions (or a strategy guide) probably started off its life in a designer’s Excel sheet.
  • Excel is great for keeping track of tasks and status, which is useful for a complicated game with lots of systems and components. If you’ve got a table with a couple hundred monsters and all their stats, it might also have an entry for whether the art for that monster is done, or whether the stats for that monster have been balanced or playtested yet.
  • Spreadsheets are good for collecting and manipulating statistics in your game. In a sports game where each player has a list of stats, are all of the teams balanced with one another? Sum or average all of the stats on the team, and you can get some idea of the overall strengths and weaknesses of each team. In a game with transitive relationships, is each game object balanced? Add up the costs and benefits in a spreadsheet.
  • You can use spreadsheets to run statistical simulations. By generating random numbers (in Excel you can use the RAND() function and press F9 to reroll), you can generate random die-rolls for things like damage within a range, many times, to see the overall range and distribution of outcomes. (Statisticians call this a “Monte Carlo” simulation, in case you were wondering.)
  • Spreadsheets help you to see causes and effects of changes in the game. By creating formulas based on specific values that you want to change, you can change one value and see what happens to the other values that depend on that one. For example, if you’re working on a Massively-Multiplayer Online RPG, you could use Excel to compute the damage-per-second of a weapon and then instantly see how that changes when you modify the base damage, accuracy, or attack speed.

Use the Rule of 2. Suppose you have some number in your game that you know is too high, but you don’t know how much. Maybe it’s just a little bit too high, or maybe it’s quite a bit off. In either case, cut it in half. Likewise, if you have a value that you know is too low, regardless of how much too low it is, double it. If you aren’t 100% sure of what the correct value is, double it or cut it in half. This is the “Rule of 2.”

At face value, this sounds rather ridiculous. If the cost of a gemstone is only 10 to 20 percent too low, what could be gained by taking the drastic measure of doubling it? In practice, there are some reasons why this works. First, you might think that it’s only slightly off, and you might be wrong; if you only make minor adjustments and the value really did need to be doubled, it will take you much iteration to get to where you needed to be in the first place.

There is a more powerful force at work, though, when applying the Rule of 2. Game design is a process of discovery. The fact is, you don’t know what the correct numbers are to balance your game; if you did, the game would be balanced already! If one of the values in your game is off, you need to discover what the correct value is, and you do this by changing the value and seeing what happens. By making a major adjustment, you will learn a great deal about the effect of this value on the game. Maybe it did only need a minor adjustment, but by doubling or halving the value, you will learn so much more about your game.

Occasionally, you will also find that by making such a large change to your game, it changes the dynamics in a way that was unexpected but (accidentally) superior to your original design.

Balancing the first-turn advantage. In turn-based games in particular, it is common for there to be a very slight advantage (or disadvantage) to going first. This is not always the case, but when it is, there are a few common techniques for compensating:

  • Rotate who the first player is. In a four-player game, for example, after each complete round (where every player has a turn), rotate the starting player to the left for the next round. In this way, the player who goes first on this round will go last on the next round. (When I was growing up, my game group used a pencil to mark the first player, so we dubbed this the “Pencil of Power” technique.)
  • Give the disadvantaged players some extra resources. For example, if the objective of the game is to score the most points by the end of the game, give each player a different number of points to start the game, with the last player having slightly more points to compensate for the disadvantage of going last.
  • Reduce the effectiveness of early turns for the first players. In a card game, maybe players typically draw four cards at the start of their turn. You could modify this so that the first player only gets to draw one card, the next player draws two, and so on until all players are drawing four.
  • For very short games, play a series of games where each player gets to go first once. This is common with card games, where a complete game is played in a series of hands.

Write down your own rules as you learn them. As you design games, you will have successes and mistakes. You will learn from both (especially your mistakes). When you find a “law” of game design or a new game balance technique, record it and review your notes periodically. Tragically, very few designers actually do this; as a result, they cannot pass on their experience to other designers, and they sometimes end up making the same mistakes on multiple games because they forget the lessons learned from earlier ones.

The Role of Balance

In my early days as a designer, I was very passionate about game balance. “A fun game is a balanced game, and a balanced game is a fun game” was my mantra. I’m sure I annoyed many of my seniors with my rantings on the imbalances in our games and why we needed to fix them immediately. Such is the privilege of youth.

While I still believe balance is important and a key skill for any game designer, I now take a more moderate line. I have seen some games that are fun in spite of being unbalanced. I’ve seen other games that are fun specifically because they are intentionally unbalanced. I have seen some games that have done astoundingly well in the marketplace, despite having egregious imbalances. These are rare and special cases, to be sure. But let this serve as a reminder to you that the techniques and concepts discussed today should never take priority over the ultimate design goals for your game. Let these techniques be your tools, not your master.

Homeplay

Your homeplay this past Monday was to arrange for a blindtest session, to be completed on or before next Thursday (August 27). Continue working on this if you have not completed it already.

Your other task, before next Monday (August 24), is to critically analyze your game with respect to game balance. Playtest the game once (either on your own or with others) with the purpose of finding balance issues – instruct everyone to intentionally seek out optimal strategies and exploit them. Play to win. Be vicious.

Then, from there, think of what systems and game objects are in need of modification (sometimes this is called tuning). Think about what techniques best fit your game. Are there transitive or intransitive relationships? Is it more suitable to balance primarily with playtesting, math, your own instincts, or some combination of the three?

Lastly, decide on the most serious game balance issue you identified in your earlier playtest. Make at least one change and playtest again (you can probably do this in the same test session). Did you fix the problem, or at least reduce its severity? Then, think about what other changes you might make, and what you will look for in future testing.

About these ads

7 Responses to “Level 16: Game Balance”

  1. Fabrizio Marcotulli Says:

    Excel is also pretty good for keeping players accounting during a game (so you don´t have to print bills), and it can be wired so it tracks buys, sells and other mundane things automatically. It can even track wins and losses in combat against various players, though that is a little more complicated.It really is like playing the game behind the scenes. However, you do need someone to be at the computer for all of this, as it is not so practical to have everyone input their data-

    I had a little experience toying with spreadsheets when trying to prototype our first game (prior to this course, as a hobby of sorts). Mostly to figure out where players were at on each round compared to where we wanted them to be, so we could adjust accordingly. It really shed some light on some aspects that made the first rounds too slow (we managed to cut playtime in half….to about 4 hours…so there is still much to do, but still)

  2. Adolfo Says:

    Yeah, Excel in a good friend.

    I used it to simulate a card DECK and shuffle.
    to do that, add a column with rand(), to generate the values and re arrange the data. Then when solo play pick the first row as if you was picked up the first card of the deck.

    • ai864 Says:

      RAND() is very useful. If you want to get even more advanced with deck shuffling, you can put a RAND() in a column next to each card, then use RANK() to give the order from smallest to largest random number (this gives a random ordering of the cards), and then VLOOKUP() using the card list and corresponding rank to create a randomized list of cards.

  3. pmigas Says:

    This blog discusses game ballancing and how a small change can make a big difference. See http://publisherperspective.blogspot.com/2009/08/playtest-notes-on-initiative-frontline.html

  4. Fabrizio Marcotulli Says:

    I guess this is coming in a little late, but David Sirlin has a series of articles/blog posts dealing with balance in multiplayer games. It starts here:

    http://www.sirlin.net/articles/balancing-multiplayer-games-part-1-definitions.html

    the series deals mostly with examples from fighting games, which I appreciate considering the most interesting fighting games can be very assimetric. He also mentions card games though, so that can draw you in if guilty gear is not your thing.

    He also has a PDF of a handout he gave for his lecture at the 2009 GDC on the same topic the articles discuss, though it is mostly a summary of the content on his blog

  5. What is Game Balance Concepts? « Game Balance Concepts Says:

    [...] summer, I’m diving deeper into the narrow topic of game balance. The topic was covered in a single lesson last year, but I think it deserves better treatment, so this year I’m looking at balance from [...]

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s


Follow

Get every new post delivered to your Inbox.

Join 143 other followers

%d bloggers like this: